University of Ghana http://ugspace.ug.edu.gh UNIVERSITY OF GHANA COLLEGE OF HUMANITIES INCOME DIVERSIFICATION, LENDING AND EFFICIENCY ON CREDIT UNION FINANCIAL PERFORMANCE IN GHANA BY BENJAMIN AMOAH (10042960) A THESIS SUBMITTED TO THE SCHOOL OF GRADUATE STUDIES IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE AWARD OF DEGREE OF DOCTOR OF PHILOSOPHY IN FINANCE DEPARTMENT OF FINANCE DECEMBER 2017 University of Ghana http://ugspace.ug.edu.gh UNIVERSITY OF GHANA COLLEGE OF HUMANITIES INCOME DIVERSIFICATION, LENDING AND EFFICIENCY ON CREDIT UNION FINANCIAL PERFORMANCE IN GHANA BY BENJAMIN AMOAH (ID. 10042960) A THESIS SUBMITTED TO THE SCHOOL OF GRADUATE STUDIES IN PARTIAL FULFILLMENT OF THE REQUIREMENT FOR THE AWARD OF DEGREE OF DOCTOR OF PHILOSOPHY IN FINANCE DEPARTMENT OF FINANCE DECEMBER 2017 i University of Ghana http://ugspace.ug.edu.gh DECLARATION I hereby declare that this thesis is my own work produced from research I carried out under supervision. This thesis has not been presented by anyone for any academic award in the University of Ghana Business School, University of Ghana or any other institution. All references made to work done by other people have been duly acknowledged. I am solely responsible for any shortcomings in this thesis. ………………………………… ……………………….. Benjamin Amoah Date (Student) ii University of Ghana http://ugspace.ug.edu.gh CERTIFICATION We hereby certify that this thesis was supervised in accordance with procedures laid down by the University of Ghana. ………………………………… ……………………….. Professor Godfred Alufar Bokpin Date (Principal Supervisor) ………………………………… ……………………….. Professor A. Q. Q. Aboagye Date (Co- Supervisor) ………………………………… ……………………….. Dr. Kwaku Ohene Asare Date (Co- Supervisor) iii University of Ghana http://ugspace.ug.edu.gh ACKNOWLEDGEMENTS I would like to express my profound appreciation to Central University, Accra for the financial support they provided that enabled me to undertake this study. An academic Endeavour such as a thesis has many individuals playing significant roles. Special mention must be made of my supervisors, Professor Godfred Alufar Bokpin, Professor A. Q. Q. Aboagye and Dr.Kwaku Ohene-Asare, who provided much guidance, support and direction throughout the preparation of the thesis. I would forever be grateful for your commitment and concern through the entire thesis process. In this regard I say God richly Bless you. I also appreciate the encouragement from Professor Donald J Smith of Boston University whose works on Credit Union and personal emails have inspired me to research on Credit Unions. My PhD finance colleagues Gloria, Haruna, George, Anthony, Myram, Josephine, Hamdiya, Able, Ebenezer, Stella, Charles Turkson, and Christian have all been good friends and provided stimulating company during the course of our studies. I say a big thank you to Ms. Edwina Ashie-Nikoi for proof reading the thesis. Miss Anita Aidoo and Richard Oyere Yirenkyi, both of CUA head office, Stella Tasa of Tema CUA chapter I say God favour you all times for your assistance in the data collection exercise. To my parents Mr Daniel Amoah and Mrs Salomey Ashong I say thank you for your encouragement. I thank my siblings, Hilda, Julie, Cynthia, Nana,Obeng and Reggie for all their support. And finally, I very much appreciate the concern, thoughtfulness and love shown by my wonderful wife, Jemima Abena Yakah-Amoah. iv University of Ghana http://ugspace.ug.edu.gh ABSTRACT Available global statistics show that credit unions have, and continue to gain more grounds as penetration rates, asset size and savings levels suggest. Yet traditional depository institutions, especially banks, still present strong competition in the areas of revenue generation, lending and efficiency for the credit union industry. Using data on 61 credit unions in Ghana over the period 2008-2014, this thesis analyses the connections between non-loan income, credit union lending and efficiency in the credit union business without leaving out banking sector development. Specifically we analyse the relationship between non-loan income and credit union performance, examine the determinants of non-loan income. Further we examine discretional and non-discretional factors on credit union lending. Finally, we probe into credit union efficiency and assess the intra and extra determinants over the sample period. We employ the random effect, Hausman-Taylor and truncated Tobit panel data models on risk adjusted return on asset and risk adjusted return on equity to assess how credit union factors influence non-loan diversification income and diversification within the liquid-financial investment activities. For credit union lending we use the fixed effect and random effect model to evaluate discretional and non-discretional factors on credit union. For our objective on cost efficiency and technical efficiency we used a two stage method, we first estimate cost efficiency using Tone’s measure in the Data Envelopment Analysis method and technical efficiency in a variable returns to scale setting for the period 2008 to 2014. For the second stage we estimate a Mixed-effects and Two-Limit Truncated v University of Ghana http://ugspace.ug.edu.gh Tobit regression to examine credit union specific, banking industry and macroeconomic conditions on efficiency. We find empirical support that non-financial income plays a role in diversifying credit union income. There exists a complementing effect on income from combined non-loan income and investment in liquid-financial asset. Our results suggest that in the case of non-financial income, size, liquidity, loan portfolio, resource usage and net worth are important. For liquid-financial investment, liquidity, age, and net interest margin play a critical role. With respect to loan portfolio, discretional factors such as size, return on equity, management quality and solvency positively associate with credit union lending business. On the other hand, factors such as loan loss, non-loan income diversification activities and lending rates negatively relate to credit union loan portfolio. For the non- discretional factors, increase in banking sector inefficiency would imply more loan business for credit unions. The results also show the presence of a non-linear inverted U- shaped relationship between credit union lending and the size of the credit union. Credit unions’ cost efficiency averaged 38.9 percent whiles average technical efficiency level was 54.4 percent. These efficiencies are mostly influenced by internal factors and the banking industry. A monopolized and inefficient banking sector does not challenge efficiency improvement in the credit union industry. We also find that technical efficiency does not necessarily translate into cost efficiency for credit unions. vi University of Ghana http://ugspace.ug.edu.gh We recommend that if management’s goal is diversification, then non-financial income is the best option to pursue against investment in liquid-financial asset, especially for smaller sized credit unions. On the other hand, if the objective is to supplement income, then management should consider liquid-financial investment irrespective of the size of the credit union. Credit union managers should monitor developments taking place in the loanable funds market as banks’ increasing overhead cost may mean a possible increase in loan demand in the credit union. Additionally, managers of credit unions should increase loan portfolio carefully so as not to experience any diseconomies of scale from the loan business. We suggest that when targeting cost efficiency, credit union managers should make technical efficiency a priority. Also credit union managers should observe and monitor the activities of the big banks, since their activities have an implication for the adoption of strategies that would help improve overall efficiency in the credit union. vii University of Ghana http://ugspace.ug.edu.gh TABLE OF CONTENTS DECLARATION ............................................................................................................. ii CERTIFICATION .......................................................................................................... iii ACKNOWLEDGEMENTS ............................................................................................ iv ABSTRACT ..................................................................................................................... v LIST OF TABLES ......................................................................................................... xii LIST OF FIGURES ...................................................................................................... xiv CHAPTER ONE .............................................................................................................. 1 INTRODUCTION ........................................................................................................... 1 1.1 Background to the Study ............................................................................................1 1.2 Research Problem ......................................................................................................7 1.3 Objectives of the Study ............................................................................................15 1.4 Research Question ....................................................................................................16 1.5 Research Hypotheses................................................................................................17 1.6 Motivation of the study ............................................................................................20 1.7 Significance of the study ..........................................................................................21 1.8 Scope of the Study....................................................................................................25 1.9 Structure of the Thesis..............................................................................................26 CHAPTER TWO ........................................................................................................... 27 LITERATURE REVIEW .............................................................................................. 27 2.1 Introduction ..............................................................................................................27 2.2.1 The Historical Development of Credit Union .......................................................28 2.2.2 Cycle of the Credit Union .....................................................................................30 2.2.3 The Credit Union movement in Ghana .................................................................31 2.3 Theoretical Models of Credit Unions .......................................................................35 viii University of Ghana http://ugspace.ug.edu.gh 2.3.1 Static theoretical framework of the Credit Union Decision Making. ...................37 2.3.2. Dynamic framework of the Credit Union Decision making. ...............................41 2.4 Theoretical Review on Income Diversification ......................................................43 2.4.1 Revenue Diversification and Financial performance. ...........................................46 2.4.2 Diversification and Agency ...................................................................................50 2.4.3 Diversification and Risk ........................................................................................53 2.5 Theoretical Review on Lending ...............................................................................56 2.5.1 Transaction Based Lending ...................................................................................57 2.5.2. Relationship Lending ...........................................................................................59 2.5.3 Macroeconomic Environment ...............................................................................60 2.5.4 Other Factors .........................................................................................................61 2.5.5 Credit Union Lending............................................................................................62 2.6.1 Theoretical review Cost Efficiency and Technical Efficiency ..............................64 2.6.2 Efficiency Empirical Review ...............................................................................69 2.7 Conclusion and Research Gap..................................................................................74 CHAPTER THREE ....................................................................................................... 79 METHODOLOGY ........................................................................................................ 79 3.1 Introduction ..............................................................................................................79 3.2 Research Design .......................................................................................................79 3.3 Population of Study ..................................................................................................81 3.4 Sampling Technique and Sample Size .....................................................................81 3.5 Data Sources .............................................................................................................82 3.6 Measuring Income Diversification ...........................................................................82 3.7 Financial Performance of the Credit Union .............................................................85 3.7.1 Income Diversification Estimation Issues .............................................................89 ix University of Ghana http://ugspace.ug.edu.gh 3.8 Credit Union Lending...............................................................................................91 3.8.1 Credit Union Lending Econometric Estimation ....................................................93 3.9 Credit Union efficiency ............................................................................................95 3.9.1 Inputs and Outputs Specification ........................................................................101 3.9.2 Efficiency Data Source........................................................................................105 3.9.3 Efficiency Econometric Estimation ....................................................................105 3.10 Integrated Performance of the Credit Union ........................................................109 CHAPTER FOUR ........................................................................................................ 111 RESULTS AND DISCUSSION .................................................................................. 111 4.1 Introduction ............................................................................................................111 4.2 Descriptive Statistics ..............................................................................................111 4.3 Income Diversification ...........................................................................................116 4.3.1 Non-Loan Income Diversification ......................................................................116 4.3.3. What drives Liquid-Financial Investment? ........................................................128 4.3.4 The case of Combined Non-Loan Income ..........................................................135 4.4 Credit Union Lending.............................................................................................139 4.4.1 Lending Empirical Results Discussion ...............................................................141 4.5 Efficiency ...............................................................................................................147 4.5.1 Cost Efficiency Analysis .....................................................................................148 4.5.2 Technical Efficiency Analysis ............................................................................151 4.5.3 Comparing Cost Efficiency and Technical Efficiency. .......................................154 4.6 Determinants of Credit Union Cost efficiency and Technical efficiency ..............158 4.6. Integrated performance of the Credit Union .........................................................171 4.7. Hypothesis Testing ................................................................................................177 CHAPTER FIVE ......................................................................................................... 179 x University of Ghana http://ugspace.ug.edu.gh SUMMARY OF FINDINGS, CONCLUSION AND RECOMMENDATION .......... 179 5.1 Introduction ............................................................................................................179 5.2 Summary of Findings .............................................................................................179 5.3 Contributions of the Study .....................................................................................182 5.4 Conclusion ..............................................................................................................183 5.5a. Recommendation ................................................................................................186 5.5 b. Policy Recommendation ....................................................................................187 5.6 Limitations and Suggestions for Further Research ................................................188 REFERENCES ............................................................................................................ 190 APPENDIX 1: NOTICE NO. BG/GOV/SEC/2017/06 ............................................... 206 APPENDIX 2A. Non-Financial Income Diversification Histogram ........................... 210 APPENDIX 2B. Liquid-Financial Investment Diversification Histogram.................. 210 APPENDIX 2.C: Combine Non-Loan Income Diversification Histogram ................. 211 APPENDIX 3.A: Tone’s Efficiency Scores Histogram .............................................. 212 APPENDIX 3.B: Technical Efficiency Score Histogram ............................................ 212 APPENDIX 4a: Scatter plots ....................................................................................... 213 APPENDIX 5.a: Non- Constant Integrated Cost Efficiency Financial Performance .. 216 APPENDIX 5.b. Non-Constant Integrated Technical Efficiency Financial Performance ....... 217 APPENDIX 6A. Sample Comprehensive Income of a Credit Union ..........................218 APPENDIX 6B. Sample Statement of Financial Position of a Credit Union .............219 xi University of Ghana http://ugspace.ug.edu.gh LIST OF TABLES Table 1.1 Descriptive Statistics on Loan for sampled Credit Unions and Banks ..............13 Table 2.1 Credit Union Chapter and Bond Type .............................................................. 33 Table 3.1: Chapter Distribution of Data set ...................................................................... 80 Table 3.2 Income Diversification and Financial Performance estimation variables. ....... 89 Table 3.3: Variable in the Lending Model. ....................................................................... 94 Table 3.4 : Efficiency and Regression Variables ............................................................ 103 Table 4.1 Combined Variables Descriptive Statistics ..................................................... 112 Table 4.2: Correlation Matrix of Income Diversification Variables............................... 115 Table: 4.3 Financial Performance, Diversification Income and Investment .................. 118 Table 4.4a: Determinants of Non-Financial Income ....................................................... 123 Table: 4.4b Non-Financial Income ................................................................................. 127 Table 4.5a: Liquid-Financial Investment .........................................................................129 Table 4.5b: Liquid- Financial Investment ........................................................................132 Table 4.6 Non-Financial Income and Liquid-Financial Investment ...............................133 Table 4.7a: Combined Non-Loan Income .......................................................................135 Table 4.7b: Combined Non-Loan Income ......................................................................137 Table 4.8: Correlation Matrix for Credit Union Lending Variables ............................... 140 Table 4.9: Lending Empirical Estimation ........................................................................143 Table 4.10. Descriptive states for variables in First Stage DEA .....................................148 Table 4.11a: Cost Efficiency Scores ...............................................................................150 Table 4.11b: Technical Efficiency Scores .......................................................................153 Table 4.12. Non-Parametric Wilcoxon Singed-Rank Test ..............................................156 Table 4.13. Paired T- test ................................................................................................157 Table 4.14a: Correlation Matrix for Cost Effieicny and Technnical Efficiency Variables ........ 159 Table 4.14b: Correlation Matrix for Cost Efficiency and Technical Efficiency Variables ........ 160 xii University of Ghana http://ugspace.ug.edu.gh Table 4.15a: Cost Efficiency Regression Result ..............................................................162 Table 4.15b: Regression Results Technical Efficiency ...................................................166 Table 4.15c: Tobit Regression Results ...........................................................................169 Table 4.16a: Integrated Cost Efficiency Financial Performance .....................................174 Table 4.16b: Integrated Technical Efficiency Financial Performance ............................176 xiii University of Ghana http://ugspace.ug.edu.gh LIST OF FIGURES Figure 1.1: Trend in Loan Income and Combined Non-loan income ................................. 9 Figure 1.2: Trend of Liquid-Financial Investment Income and Non-financial Income ... 10 Figure 1.3: Credit Union Loan growth and Bank Loan growth . ..................................... 11 Figure 1.4. Conceptualisation of Financial performance, Efficiency, Lending and Non Loan Income Diversification. ........................................................................................... 20 Figure 2.1: Credit Union Integrated performance model.................................................. 77 Figure 4.1: Growth Trend in Tones Efficiency and Technical Efficiency ..................... 155 Figure 5.1 Established relationship in the credit union performance Nexus ..................177 xiv University of Ghana http://ugspace.ug.edu.gh xv University of Ghana http://ugspace.ug.edu.gh CHAPTER ONE INTRODUCTION 1.1 Background to the Study Financial institutions were previously presented as relying solely on one dominant source of income. For banks in particular this was income generated from loans; however, this situation has been changing, a condition mainly attributable to deregulation of the financial services industry with its attendant competitions. The competitive nature of the financial services industry has created a natural desire for deposit taking financial institutions to couple new income avenues with their traditional sources. This additional income comes from harnessing opportunities which were hitherto not permitted by regulators of depository financial institutions. Although these additional income generating activities add to income and stabilize returns, they also expose the financial institutions to new forms of risk. Landskroner, Ruthenberg and Zaken (2005) argue that to meet their profit goals, financial institutions can use diversification to improve profits and galvanise operational efficiency, especially if the scale and scope of operations are to increase. Notable among these financial institutions that offer financial services are credit unions. Berthoud and Hinton (1989) referred to credit unions as co-operative societies which offer loans to their members out of the pool of savings that are built up through member deposit. In meeting its cooperative objective credit unions provide their members 1 University of Ghana http://ugspace.ug.edu.gh financial services that are also offered by other financial institutions, namely banks, thrifts institutions, and finance companies, among others. McKillop and Wilson (2011) hold the view that credit unions are self-help cooperative financial organizations geared towards attaining the economic and social goals of members and wider local communities. The membership of the union is defined by a common bond, this bond can be the same employer; in this case a workplace credit union, belonging to the same association or society; a parish or association based credit union, or living within the same community; which is a community based credit union or residential credit union. According to Ferguson and McKillop (1997), credit unions’ main objective has been the promotion of thrift which is directed at fulfilling human and social needs; thus credit unions are not formed to make profits but rather to maximise the benefits to their members (self-help). The credit union is governed by its members, members elect (from within that membership), and during initial stage of development unpaid volunteer officers and directors who establish the policies under which the credit union operates. Therefore, in comparison to banks that can to a large extent differentiate owners from customers, credit unions, may be regarded as financial cooperatives which have owners (or members) who are depositors, borrowers and claimants on the residual income, these members determine the composition of the board of directors (Jones, Kalmi, & Kauhanen, 2012). From these presentations, the conclusion that can be drawn is that credit union is the purest of all cooperatives; operated by members, the clients are the same members, for the members benefit. 2 University of Ghana http://ugspace.ug.edu.gh As a financial institution, the credit union is somewhat of a paradox. It cannot be classified as a purely classical economic firm that seeks to maximize profit. The credit union’s cooperative nature complicates any decision-making that upholds profit making as its sole objective as it cannot then simultaneously maintain its social objective mission to provide financial services to its members at lesser cost than other competing financial institutions. Credit unions operate in a vigorous competitive environment made up of banks, savings and loans institutions, microfinance institutions which all offer similar products and services aimed at the same borrower with little variation in product features. As competition tightens revenue markets for financial institutions, the case for smaller financial institutions’ such as the credit unions’, revenue protection, strategies for increasing revenues and value creation for owners becomes imperative. What makes the case of credit union peculiar in this competitive market environment is that operationally, credit union grant loans at low rates to borrower oriented owners and at the same time pay high dividend and high deposit rates to saver oriented owners. This borrower-saver conflict in the credit union business exerts pressure on the management to exploit avenues that would minimize this conflict in their resource management. To effectively do this, management would have to consider increasing revenues of the credit union without jeopardizing its loan creating ability. It is clear, then, that non-loan income would have a significant role to play in achieving this income objective. Inferring from Smith’s 1984 view that, income from non-loan activities can be used to subsidize 3 University of Ghana http://ugspace.ug.edu.gh the interest charged on loans extended to borrower oriented members of the credit union. In the credit union setting, income from non-loan activities is in two categories, the first is non-financial income: which includes, entrance fees, commission income and fees from business activities, selling of t-shirts, selling of building materials, income from operating pharmaceutical shops, guest centers, selling of laptops among others. The second category is liquid financial income: this is interest income from investing in financial instrument including, treasury bills, money market mutual fund, credit union central credit facility, time deposit among others. The summation of these two categories of income is the combined non-loan income of the credit union. Again income from non- loan activities is used by the managers of credit unions to augment the higher comparable market interest paid on deposit paid to saver oriented members of the credit union. Operationally, credit unions generate income by granting loans to members and invest excess reserves into interest-earning asset such as treasury bills, commercial paper and other money market instrument available on the financial markets, as well as engaging in non-financial income transactions aside loans. In granting loans, credit unions compete with other financial institutions, and to keep this business going, they must offer loans at lower rates. Although the loan demand in credit union is internal in that it is the owner-borrowers who request for loans, the decision to grant the loan is not solely influenced by credit union firm specific factors or discretional factors, in that other non-discretional factors, 4 University of Ghana http://ugspace.ug.edu.gh that is economic factors and economic agents especially banks indirectly influence credit union loan making. Also the manager of the credit union must be prepared to pay more than average market return to attract deposits away from competitors. Charging higher lending rates would result in a situation where the credit union would report lower loan income as borrowers would prefer to source for loans from competing financial institutions (Smith,1984, Rubin, Overstreet, Beling & Rajaratnam 2013). This possible lower loan income would not arguer well for saver oriented members of the credit union, who join the credit union for high deposit. For members to continue to do business with the credit union requires that benefits from the credit union exceed benefits from business transactions with other financial institutions. In the absence of this the member-owners of these credit unions would see no reason to patronise the credit union. The obvious effect of this, of course, would be that the credit union would cease to exist as competing financial institutions would take over the markets they serve. However, other members join the credit unions for non- pecuniary reasons such as waiting time, convenience, courtesy, and information disclosure among other as noted by Smith (1984). One way of achieving this survival objective is for credit unions’ management to make efficient use of resources allotted them by the owners of the union. The efficiency requirement in the management of credit union cannot be downplayed because of the peculiar nature of credit unions. In the credit union setting as already noted, business is conducted with owners and for owners. This business comes mainly in 5 University of Ghana http://ugspace.ug.edu.gh the form of owners who borrow from the credit union with the expectation of paying lower than market rates on loans and owners who save with the credit union with the objective of receiving higher deposit rates than market rates paid on deposits at competing financial institutions. This presents a natural margin squeeze which calls for a good control over expenses if the value creation for the owners of the credit union is to stay intact (Taylor, 1977). The inability to control expenses might mean managers have to report declines in financial performance to credit union owners. From the foregoing issues, one can say that non-loan income would continue to play a key role in the income basket of credit unions. Lending would continue to be the major source of income for the credit union. Efficiency would be the main channel of value creation in the credit union setting. In the presence of competition on the loanable funds market, credit unions’ lending business juxtaposed with efficiency would continue to receive research attention. The theoretical, varying empirical conclusions and sometimes scanty research evidence from developed economies (including the works of (Taylor, 1977; Smith,1984 & 1985; Ferguson & McKillop 1997; Esho, Kofman & Sharpe, 2005; Worthington, 2000; McKillop, Glass, & Ward 2005; Goddard, McKillop & Wilson 2008; Rubin, Overstreet, Beling & Rajaratnam 2013; McKillop, Glass & Ferguson 2002; Malikov, Zhao & Kumbhakar,2017; among others) on credit unions in respect to income diversification, lending and efficiency, as well as the little research attention given from emerging 6 University of Ghana http://ugspace.ug.edu.gh economies (Adusei’s 2012 study on Ghana and Mathuva’s 2016 study on Kenya being exceptions), creates a natural space for further credit union studies. 1.2 Research Problem The competitive nature of financial institutions market as a result of deregulation would imply proactive strategies at augmenting income within regulatory permits for managers of credit unions. Charging owner-borrowers low loan rates and paying owner-savers high deposit rates compared to the market results in a condition called “rate squeeze” according to Taylor (1977). In order not to subsidize loans to borrower oriented owners at the expense of the saver oriented owners, credit union managers would have to readily consider other sources of income aside loan income to increase performance in the interest of the savers, being mindful of the risk inherent in the non-loan income generating activities. For the United States, Goddard, McKillop and Wilson (2008) have shown that larger credit unions have improved their unadjusted returns, albeit assuming more risk by diversifying more rapidly into non-financial income activities. Also studying credit unions in the United States, Malikov, Zhao and Kumbhakar (2017) conclude that 27 to 91 percent of credit unions that have diversified exhibit the benefits of high levels of economies of diversification; furthermore there exists a presence of non-negligible economies of diversification of financial services in the credit union industry. 7 University of Ghana http://ugspace.ug.edu.gh Damankah, Anku-Tsede and Amankwaa’s 2014 study of small sized financial institutions in this case banks in Ghana shows that small banks are more likely to engage in non- interest earning activities than the big banks. For banks that non-interest income is a high component of total income; higher interest income, customer deposit, exposure to risk and liquidity are common factors that influence non-interest income, the central bank’s prime rates also positively affects banks’ non-interest income generations (Damankah, Anku-Tsede & Amankwaa’s (2014). Adusei (2015) recommends that in Ghana rural and community banks should invest more funds in short and long term securities and that diversification of income increases financial performance of rural banks. Adusei adds that shareholders would prefer this move because these short term and long term investment increase profit without making a case for the risk that is associated with non-loan income, however, warns that rural banks’ over concentration on non-loan activities would be at variance to the very principle of establishing rural banks, which is to provide credit facilities to alleviate rural poverty. Empirical data on credit unions sampled for this study and displayed in Figure 1.1 suggests that about 70 percent of credit unions’ income derives from loans while about 30 percent of their income comes from non-loan sources. 8 University of Ghana http://ugspace.ug.edu.gh Loan Income to Total Income Combined Non Loan income to Total Income Figure 1.1: Trend in Loan Income and Combined Non-loan income The question is why would managers of credit unions consider the 30 percent of income from non-loan sources which is invariably riskier, rather than the 70 percent loan income which is of lower risk because of the consignor effect? Attempts to answer this question would be incomplete without referring to the fact that in the credit union income scenario, non-loan income comes to the fore only in the presence of excess loanable funds over loan demand as loan default in the credit union business is not common compared to banks. Additionally, the existence of a common bond that defines members invariably reduces information asymmetry problems that are typical in the lending business. The 70 percent of funds extended as loans are to members, hence lower rates compared to the average lending rates on the loanable market. The remaining 30 percent of estimated loanable 9 University of Ghana http://ugspace.ug.edu.gh amounts would have to be invested to shore up income to pay appreciable deposit rates and dividends expected by depositors if they are to keep doing business with credit unions. Liquid Financial Investment Income to Non –Loan Income Non – Financial Income to Non- Loan Income Figure 1.2: Trend of Liquid-Financial Investment Income and Non-financial Income From Figure 1.2 we see that within the non-loan income group there exists a complete substitutability of this income source, as an increase in liquid-finance investment income associates with a decline in the credit union’s non-financial income source and vice versa. This two non-loan income is presented in a sample credit union statement of comprehensive income and Statement of financial position presented in Appendix 6a and Appendix 6b. 10 University of Ghana http://ugspace.ug.edu.gh Figure 1.3 illustrates data on sampled banks and credit unions sampled for this study, the evidence show that during the period 2008 to 2014 growth in credit union loans have been on the increase on the whole compared to a more stable loan growth in the banking sector. It also be seen that both institutions have experiences increase in loan growth for during the period 2008 to 2014. What is also clear is that there was a higher variability in credit union loans compared to bank loans extended. This can be attributed to the size of banks and the market that banks serve which makes smoothening of loans extension possible compared to the restrictive market at the disposal of credit unions. Figure 1.3: Credit Union Loan growth and Bank Loan growth . Source: Author’s processed data from CUA of Ghana and Ghana Association of Bankers. 11 University of Ghana http://ugspace.ug.edu.gh From Figure 1.3, we see that credit union loan operation mimics the operation of bank loan extension. In Table 1.1 the descriptive statistics on the sampled credit union loan and bank loan is presented, this gives a snapshot into loans extend by credit unions, compared to loans created by banks for the period 2008 to 2014. 12 University of Ghana http://ugspace.ug.edu.gh Table 1.1 Descriptive Statistics on Loan for sampled Credit Unions and Banks Loans Mean Std. Dev. Min Max CU Avg Loans Ghc 426,143.50 326,440.10 147,331.70 962,046.70 Bk Avg Loans Ghc 566,729.20 312,758.50 275,339.20 1,161,508.00 Source: Authors’ estimations form data provided by CUA and Ghana Association of Bankers Credit union loans have displayed more variability as seen in the standard deviation of Ghc326,440.10 compared, with bank loan variability of Ghc312,758.50. Since loan income makes up about 70 percent of credit unions’ income, managers of credit unions are confronted on a daily basis with a situation where they have to deal with major internal income generating activity; loans which is greatly influenced by external factors especially banking sector activities which the management of the credit union can only adapt to. In a lending and deposit mobilization industry where banks tend to lead and drive activities, efficiency in production of financial services in the credit union is critical if they are to stay relevant for their owners. The inability to control expenses on the part of the credit union would mean management would report declines in financial performance to credit union owners. The management of the credit union is therefore presented with a situation where they have to continuously work on improving efficiency both from within, through their production process, and from without, managing input at prices not under their control while being unable to pass on inefficiency as banks do by charging high prices to customers for services rendered. Cost efficiency is about being the least 13 University of Ghana http://ugspace.ug.edu.gh cost producer whiles technical efficiency concerns producing outputs by avoiding wastage of resources. In this efficiency pursuit, should technical efficiency precede cost efficiency? Which of the two should be a preoccupation for the management of the credit union as technical efficiency in most cases does not necessarily translate into cost efficiency. Further more efficiency, be it cost efficiency or technical, cannot be fully appreciated by the management of the credit union without understanding how external factors can influence efficiency, hence an additional responsibility on management is monitoring and responding to external development if efficiency is to be improved in the credit union financial service provision and the overall value creation for owners. Available information from the Bank of Ghana (BoG), on the financial services industry is that as at November 2017, Ghana had 35 Licensed Banks and 4 Licensed Representative Offices, 71 Specialized Depository Institutions, 140 Rural and Community Banks, 417 Forex Bureaus, over 300 Microfinance Companies and over 435 Credit Unions. In 2017, UT and Capital Bank had their banking licenses withdrawn by BoG for deficient capital, also 70 Microfinance companies and one money lending company had their licenses revoked for their efficiencies and the threat that their existence posed to the entire depository industry. In the midst of these developments, the credit union industry has been relatively stable. The stable condition of credit unions makes an inquiry into 14 University of Ghana http://ugspace.ug.edu.gh their operations that is Income Diversification, Lending and Efficiency, important and interesting. 1.3 Objectives of the Study The general objective is to evaluate the nature of relationship that exists between non- loan income, credit union lending and efficiency in credit unions and how they influence financial performance of credit unions. Specifically, this thesis seeks to: 1. assess how credit union specific characteristics (size, cost to income, liquidity, income diversification among others) impact financial performance; 2. evaluate the effects of credit union specific factors on non-loan income; 3. analyse the effects of discretional and non-discretional factors on credit union loan portfolio; 4. estimate and discuss cost efficiency and technical efficiency levels in credit unions; 5. discuss the interrelation between cost efficiency and technical efficiency in the sampled credit union; 6. empirically examine the determinants of credit union cost efficiency and technical efficiency; 7. empirically, assess the determinants of credit union financial performance in an integrated model. 15 University of Ghana http://ugspace.ug.edu.gh 1.4 Research Question This thesis broadly sets out to offer theoretical and empirical evidence on non-loan income diversification, lending and efficiency of credit unions as well as credit unions’ financial performance. In addition the thesis attempts to answer the following key questions: i. What is the relationship between liquid-financial investment income, non-financial income and combined non-loan income and credit union financial performance in Ghana? ii. How does credit union specific factors and banking sector development relate to the three types of diversification income (non-financial income, liquid financial investment and combined non-loan income) of credit unions in Ghana? iii. What credit union discretional and non-discretional factors influence credit union lending in Ghana? iv. What is the cost efficiency and technical efficiency levels of credit unions in Ghana? v. What intra credit union factors (size, liquidity, net worth, among others) and extra factors (banking sector development and macroeconomic variables) influence efficiency of credit unions in Ghana? vi. How, in an integrated model, do non-loan income, credit union lending, and efficiency associate with financial performance (Risk adjusted return on asset and Risk adjusted return on equity)? 16 University of Ghana http://ugspace.ug.edu.gh 1.5 Research Hypotheses From the works of Smith (1984), Rubin, Overstreet, Beling and Rajaratnam (2013) together with other empirical works such as Esho, Kofman and Sharpe (2005), Worthington (2000), McKillop, Glass, and Ward (2005), Goddard, McKillop and Wilson (2008), McKillop, Glass and Ferguson (2002) Malikov, Zhao and Kumbhakar (2017) and Adusei (2012) among others, we hypothesize the following with respect to non-loan income of credit unions in line with the research question. H1: Non-financial income has a positive effect on credit union financial performance. H2: Income from liquid-financial investment has a positive effect on credit union financial performance. H3: Combined non-loan income has a positive effect on credit union financial performance. H4: Combined non-loan income has a negative effect on credit union loan portfolio. With support from the efficiency works of Worthington’s 1998 and 2000 studies in Australian on non-bank financial institutions including credit unions, and Esho’s (2001) efficiency study on a sample of Australian credit unions, we proffer the following hypotheses on efficiency in credit unions. H5: Combined non-loan income has a negative effect on credit union cost efficiency. H6: Combined non-loan income has a negative effect on credit union technical efficiency. 17 University of Ghana http://ugspace.ug.edu.gh We hypothesize the following relationship in relation to credit union risk adjusted financial performance, following McKillop, Glass, and Ward (2005), Goddard, McKillop & Wilson (2008). H7: Non-loan income, loan portfolio, cost efficiency and technical efficiency have significant relationship with risk adjusted financial performance. From the theoretical models of Smith (1984 & 1988) and Rubin, Overstreet, Beling and Rajaratnam (2013) credit unions maximize the benefit of both borrowers and savers. For borrowers this benefit is seen in the lower loan rates, compared to market loan rates, Bauer (2007), whiles for the saver this is in the form of higher deposit rates compared to market deposit rates. Loan income as presented in figure 1.1 accounts for about 70% of income in the credit union, whiles income from non-loan that is non-financial income and liquid-financial investment income makes up the remaining 30%. In creating these outputs namely, loan income and non-loan income, the credit union would have to efficiently deploy inputs to create these outputs. This implies that efficiency is critical in the credit union benefit creation process. Credit unions are financial cooperatives, not-for-profit organised to meet the needs of their members from Goddard, McKillop & Wilson (2008), performance is important for owners of the credit union, this can be measured on a risk adjusted and unadjusted return basis. The performance emanates mainly from the loan income and from the non-loan income sources. Based on the problem statement and the hypothesis, a conceptualization of the major themes of this thesis developed in Figure 1.4. 18 University of Ghana http://ugspace.ug.edu.gh 19 University of Ghana http://ugspace.ug.edu.gh Financial Performance Efficiency Loan Income Non-Loan Income Liquid- Financial Investment Non-Financial Income Figure 1.4. Conceptualisation of Financial performance, Efficiency, Lending and Non Loan Income Diversification. Source: Author’s Conceptualisation (2017). 1.6 Motivation of the study The 2014 statistical report of the World Council of Credit Unions (WOCCU) show that there was 57,000 credit unions operating in 105 Countries on six, serving 217 million people compared to the 2015 statistical report of WOCCU the number of credit unions increased to 60,500 in 109 countries on six continents, serving 223 million people around the globe. From all indications this statistic is expected to increase in the years to come. The possible reasons for this increasing trend and popularity of credit unions include the fact that being owner managed with business transactions mainly within and among members with members savings as collateral provides there is a safety net for many a member of the credit union. This again provides as provides fertile ground for research as the data trend from the WOCCU indicates that for the foreseeable future consumers will 20 University of Ghana http://ugspace.ug.edu.gh turn to credit unions for the needed safety of financial transactions as mainstream financial institution such as the banks have on many occasions failed to give, the most recent case in point being the 2007 financial crises that rocked most banks in developed economies, especially the United States and the United Kingdom. Also Myers, Cato and Jones (2012) proffer that credit unions in many rural areas and deprived communities have served as the main stream financial institutions providing financial service as many a bank have overlooked the financial services needs of these communities on cost grounds and for strategic reasons such as consolidation purposes. It must also be added that in the big cities, credit unions have also maintained a formidable presence in the provision of financial services. We can conclude from this that the credit union by its nature is closer to a critical mass of the populace than banks that can choose and pick where to operate. A study on this critical economic agent is therefore important as in some cases the credit union is seen as a second option for the provision of financial services and products when in fact their existence gives members (customers) an opportunity to exercise a direct control over the activities of a financial institution. 1.7 Significance of the study This thesis makes a few important contributions to the literature. It fills the gap in the existing literature by focusing on the factors that impact on credit union non-loan income with additional emphasis on non-financial transactions in a developing market context, a 21 University of Ghana http://ugspace.ug.edu.gh shift from Smith 1984 financial transaction theory based work and the extant empirical evidence that exist on diversification and its impacts on performance of banks (DeYoung and Rice, 2004; Busch and Kick, 2009; Sanya and Wolfe, 2011; Trujillon-ponce, 2013; Brighi and Venturelli, 2014; Doumpos, Gaganis and Pasiouras,2016; Nguyen, Nghiem and Nghiem, 2016) and credit unions in particular (Esho, Kofman and Sharpe, 2005; Goddard, McKillop & Wilson, 2008). In addition, this thesis contributes to the discussion of diversification in credit union operations by considering the causal agents of diversification income for credit unions; understanding these factors is particularly crucial in the wake of increasing reliance on non-loan income as a source of additional revenue for credit unions. As mentioned earlier, the central issue facing the manager of the credit union is how to keep meeting loan requests from borrowers while fully aware that this borrower could request loans from other providers. This suggests that external forces do indirectly influence credit unions’ loan making process. Since loan income makes up about 70 percent of credit unions’ income, the manager of the credit union is presented with a case where they have to deal with a major internal income generating activities for which the external factors greatly influence, hence understanding the dynamics surrounding lending in the credit union is paramount, for the management of credit unions. This study with a focus on credit union lending in Ghana, differs from Adusei’s study (2012) on credit union savings which concludes that credit risk, asset (size) and female 22 University of Ghana http://ugspace.ug.edu.gh membership are significant determinants of credit union savings. In contrast, Adusei (2012) argues that credit unions in Ghana could maximize their savings if they target their savings mobilisation campaigns at income-earning men with higher marginal propensity to save as well as adoption effective and efficient asset management policy. By focusing on loan to asset we assess how this lending business is influenced by the banking sector of the financial market, a major competitor and threat to the credit union loan business and proffer suggestions on how credit union can survive this unending threat that come from banks. The management of banks and the management of credit unions in particular stand to benefit form this thesis, as to how the competitive strategy of each entity would have to be. This thesis contributes to the debate on cost efficiency and technical efficiency in a finance cooperative setting. It empirically assesses the heterogeneous ability of credit unions in accessing input at different prices on efficiency. The thesis also examines technical efficiency, an empirical gap not given ample attention in the credit union literature, by researchers. Methodologically, using the Tone measure of efficiency is a new addition to our empirical understanding to efficiency in the credit union literature. Moreover, by including the banking sector development in this study on credit union cost efficiency and technical efficiency, the impact of the “no boundary effects” in terms of conducting business with owners and non-owners in the case of banks operations on credit unions’ efficiency is carefully examined. This should be of interest to the Bank of Ghana as it continues to reform the financial services industry in Ghana. 23 University of Ghana http://ugspace.ug.edu.gh Our integrated performance model to appraise credit union financial performance in Ghana is a clear departure from Goddard, McKillop and Wilson (2008) which focused on diversification and financial performance of US credit unions, using the traditional measure of non-loan income to total revenue and not measuring efficiency from a parametric or non-parametric approach as this thesis empirically presents. The use of diversification index in the credit union setting in the integrated model is a new addition to the empirical evidence on credit union financial performance in the credit union literature. This is a departure from the traditional approach to the diversification-financial performance nexus which would be of interest to future credit union researchers. By focusing on Ghanaian credit union industry, this study provides empirical evidence on non-loan income diversification, determinants of lending, cost and technical efficiency and how all these in an integrated model drive risk adjusted financial performance from a developing economy context. This fills a gap in the existing literature which is dominated by studies from developed economies. Finally, the findings have managerial and policy implications in that, this study highlight some operational issues regarding non-loan income diversification, lending, cost effciency and technical efficiency that credit union managers may have to consider in their attempts to improve efficiency. Some of these industry wide policy interventions which could help sustain cooperative businesses are also presented. 24 University of Ghana http://ugspace.ug.edu.gh 1.8 Scope of the Study This thesis focuses on the credit union income diversification and to be specific, non-loan income, credit union lending that is loan extended out of total asset at the disposal of the management of the credit union. This research further considers cost efficiency and technical efficiency of credit unions. The choice of credit unions as a unit of study was largely influenced by the fact that studies on credit unions from emerging markets are very scanty compared to evidence available from developed economies. The sample of 61 credit unions and sampled period of 2008 to 2014 was also heavily influenced by the availability of data although the initial aim was to study all credit unions operating under the umbrella body of CUA. In arriving at the sample number of credit unions, the study relied on published annual reports submitted at the CUA head office in Accra, with further attempts to supplement the data from the CUA Tema and Accra regional offices due to data limitation and unavailability of data on some credit unions, especially those operating outside Accra. The study also eliminated credit unions that did not have that data that was suitable for its regression estimations or was missing that was relevant to our stated objectives. The study ended up with 61 credit unions for a 7 year period that is 2008 to 2014, providing annual observation data for 427 credit unions. As a significant number of the credit unions in our sample are from Ashanti, Brong Ahafo and Greater Accra chapters; and these chapters have the large sized credit unions and the highest number of credit unions licensed by CUA. We expect that the results and 25 University of Ghana http://ugspace.ug.edu.gh conclusion from this study would to a large extent reflect the situation of the credit union industry in Ghana. 1.9 Structure of the Thesis This thesis is organized into five chapters. Chapter one, the introduction, provides the background to the study, the problem statement, research questions, research hypotheses, research objectives, and significance and scope of the study. Chapter two, the literature review, builds a theoretical case for the credit union and reviews relevant theoretical and empirical literature related to income diversification, lending, efficiency and financial performance of credit unions. Chapter three covers the various methodologies employed to achieve the objectives of this study. Chapter four presents the results obtained and discussion analyses them. Chapter five provides the summary of findings, conclusions, contributions of the study, policy recommendations, as well as limitations and suggestions for further research. 26 University of Ghana http://ugspace.ug.edu.gh CHAPTER TWO LITERATURE REVIEW 2.1 Introduction The literature review section of this thesis reviews and appraises studies on credit union in terms of theories, models and empirical works related to the broad objectives of this study. In line with the objectives of the thesis, we focus our review in sections dealing with diversification, lending, efficiency and performance related studies and their respective determinants in the credit union setting. The literature review begins with some theoretical postulations advanced for setting up the credit union by focusing on the work of Smith (1984) and Rubin, Overstreet, Beling & Rajaratnam (2013). The strategy adopted is that each section is reviewed theoretically or conceptually and empirically relying on recent and available empirical literature discovered through literature search from the University of Ghana Balme Library Electronic Resources. Section 2.2 is dedicated to the historical development of credit union and the credit union movement in Ghana. Section 2.3 looks at the theories that exist on the objective of setting up the credit union, with particular emphasis on the work of Smith 1984 and Rubin et al (2013). Section 2.4 is dedicated to income diversification, the reasons for the pursuit of non-loan income in the financial institutions literature, drawing much on the extant 27 University of Ghana http://ugspace.ug.edu.gh evidence from banking studies and those available on credit union studies. In section 2.5 we consider lending, and in particular loans created out of the asset of credit unions; this section also discusses the factors that influence lending in the credit union literature. Section 2.6 deals with efficiency with special interest in cost and technical efficiency. The literature review ends with section 2.7 a conclusion on the evidences available from empirical works as well as the presentation of identified research gaps. 2.2.1 The Historical Development of Credit Union According to Fairbain (1994) the history of cooperative credit unions can be traced to the documented consumer cooperative group, Fenwick Weavers’ Society, founded in 1769 by a group of weavers. This society was formed to see to the interest of weavers who were members of this society. Also from the study of Feinberg, and Rahman, (2006, p.12), the cooperative ideals in the 19th central can be traced to Great Britain when in 1844, from the experiments of Robert Owen at New Lanark Mills where equality and cooperation was the main passion of Robert Owen, where he achieved marginal success. Although his New Lanark Mills socialist and cooperative ideology failed, the influence of this experiment proved to be widespread and enduring (Feinberg & Rahman, 2006). Another view is that in 1844, the Rochdale Society of Equitable Pioneers established a cooperative in which members subscribe to shares to raise capital to buy inventory in 28 University of Ghana http://ugspace.ug.edu.gh bulk from wholesalers at prices less than retail price. These inventories were sold to their members at a surplus for the society. This surplus was share proportional to share ownership by members at the end of the year. The Rochdale Society adopted the ‘Rochdale Principles’ which included voluntary and open membership, democratic member control, member economic participation, autonomy and independence, education, training and information, cooperation among cooperatives, concern for community. These principles have over the years become the main principles on which modern day cooperative including the credit unions operates (Mckillop and Wilson, 2011; Moody & Fine, 1971). Despite the efforts of the leaders of these cooperatives, these societies were not so successful. From Germany two groups of successful cooperative credit unions emerged, the first group was the cooperative credit union started by Hermann Schulze-Delitzsch in 1852. The Hermann Schulze-Delitzsch credit union mobilized saving and credit from traders, shop owners and artisans in urban areas and stressed paying dividends to members, and applied limited liability principle (Mavenga, 2010). The second group was formed in 1864, when Friedrich Wilhelm Raiffeisen, a mayor in the western Rhineland area, formed the first rural credit cooperative and focused on helping rural farmers (Hannan, 2003). Whiles Hermann Schulze-Delitzsch was interested in the areas of urban cooperatives, the concern for Friedrich Wilhelm Raiffeisen was in rural cooperatives. It is from the success stories of these early Germany cooperatives that the concept of cooperatives financial institutions reached other European countries like the Italy’s 29 University of Ghana http://ugspace.ug.edu.gh People’s Bank with a limited liability ownership form, heavily influenced by the works of Italian scholar Luigi Luzzatti who have had some insight into the German concept of cooperative from Isbister (1994). The cooperative concept then caught up in Canada in the 20th century according to Frame et al (2001, 31). This was inspired by Alphonse Desjardins, who was moved by the plight of the poor who were heavily exploited by lenders. Similar to the people’s bank concept in Italy, Desjardins established the Caisse Populaire in Levi, Quebec. The Caisse Populaire combined Catholic prohibition of usury and the Quebec political and religious philosophy of „la survivance in managing the new Canadian form of cooperatives. The cooperative idea then caught up in the in America, when Edward Filene, formed a savings and loans association for his employees which later became the Filene Employee’s Credit Union. Later Roy Bergengren popularized the credit union movement in the US having been associated with the Filene credit union and the formed the Credit Union National Association (CUNA) and CUNA Insurance. He held the view that the credit union could be used to improved the lot of people. 2.2.2 Cycle of the Credit Union Three stages of credit union development exist according to McKillop, Ferguson, and O'Rourke (1997), these are the “nascent stage” characterized by self-help concept and strong social reason for existence, credit union at the nascent stage are small in size, tight 30 University of Ghana http://ugspace.ug.edu.gh common bond exist between members and voluntarism support services provided by members for the union. The “transitional” stage is marked by the desire to change, relax the cooperative principle and a strong wish to cut down cost as the scale of operations increase supported by growth in credit union size. Competition from other financial institution tightens up the market for the credit union industry in this stage. The last stage of development is the “mature” stage, this is when credit union assumes a lot of the features of a main stream profit oriented financial institution, the asset size becomes large and operations of the union are professionally managed with less voluntarism. The range of products on offer also increases as a results of diversification, services are centralised to help to cut down operational cost and the common bond feature is less emphasized as membership continues to grow. 2.2.3 The Credit Union movement in Ghana Credit Union history in Africa is traced to Jirapa in the Upper West Region of Ghana, when in September 1955, the first credit union was introduced by Reverend Father John McNulty an Irish Canadian, and a catholic priest. The reason for setting up this credit union was “To mobilize savings for development, to change the traditional way of saving, to encourage the people in the area and elsewhere to develop themselves, economically and to be self-sustainable, and to inculcate the spirit of co-operation among the people” (CUA News Letter 1998). 31 University of Ghana http://ugspace.ug.edu.gh To give the movement a national backing, information from CUA Ghana has it that in 1960, Pope John XXIII appointed Bishop Dery as the Bishop of Wa, he encouraged the formation of Credit Unions in all the Parishes. Among them were Nandom, Kaleo, Ko, Daffiama, Wa, Lawra and Tumu. The Bishop gathered courage at that time and met the President Dr. Francis Kwame Nkrumah to discuss issues relating to the Credit Unions movement in Ghana. Before the 1955 credit union success, the Department of Cooperatives in 1920 realized the need for popular credit and savings facilities in Ghana. The interest of these cooperatives then was for government employees and some cocoa farmers to harness financial resource for the benefit of their members. These cooperatives had dominance and physical presence in the south, however these cooperatives where short lived because of poor management and later on some politicization that characterized these economic groups. In 1961, the Convention People Party (CPP) under the first president of Ghana Dr Kwame Nkrumah dissolved the Department of Cooperatives and the cooperative banks because the government did not have political interest in these Savings and Loan societies. In 1967, the Credit unions in the North were united in a chapter because the white Fathers had been advisors to all of them and Credit unions in the south also felt a need for joining together for training programs and exchange of experience. 32 University of Ghana http://ugspace.ug.edu.gh Following a meeting sponsored by CUMA International in Lesotho in January, 1968 the idea of a National Association in Ghana was conceived. A follow-up conference was held in April in Tamale the same year giving birth to the Ghana National Union and Thrift Association, the forerunner of the Ghana Cooperative Credit Unions Association (CUA) Limited. The duties of CUA limited were to promote, educate, organize and support the Credit Union Movement nationally and internationally. Currently the credit union industry in Ghana is divided into 11 operational Chapters under the umbrella body of CUA. As at July 2017, the number of credit union licensed by CUA is 435. Three types of credit unions are registered to operate in Ghana, namely parish based, workplace based and community based credit unions. As at the end of the year 2017, CUA had registered about 435 credit unions. The breakdown of the credit unions are: 150 community based, 137 parish based, 148 workplace based and 1 study group. Table 2.1 gives a detail break down of the 435 credit unions in Ghana. Table 2.1 Credit Union Chapter and Bond Type Chapter Community Parish Workplace Total Ashanti 22 40 14 76 Brong Ahafo 25 2 8 35 Central 11 3 22 36 Eastern 16 3 16 35 Greater Accra 21 62 35 118 Northern 13 4 3 20 Tema 16 13 14 43 Upper East 2 3 9 14 Upper West 6 4 2 12 Volta 4 0 16 20 Western 14 3 9 26 Total 150 137 148 435 Source; Author’s compilation for CUA Ghana. 2017 33 University of Ghana http://ugspace.ug.edu.gh It is evident in Table 2.1 that credit unions in Ghana in terms of types are fairly represented in the case of bond type. From this same table it is clear that the Greater Accra chapter has highest number of credit unions, followed by the Ashanti chapter. CUA is the body mandated to register and regulate the activities of credit unions in Ghana. CUA, among others, is a partner of the World Council for Credit Unions (WOCCU) and African Confederation of Co-operative Savings and Credit Association (ACCOSCA). As far back as 2001, Ofei, noted that CUA has on the average chuck some admirable success in the mobilization of financial resources but cautions that CUA was not fully financially self-sufficient. For regulatory reason, credit unions are classified by the Bank of Ghana to be in the microfinance category of the financial institution industry. Credit union operation is in accordance with the Non-Bank Financial Institutions Act, 2008 (Act 774) and Co- operative Credit Union Regulations, 2015 (L.I. 2225). There exist three broad types of microfinance institutions operating according to the Bank of Ghana. These are: 1. Formal suppliers of microfinance (i.e. rural and community banks, savings and loans companies, commercial banks) 2. Semi-formal suppliers of microfinance (i.e. credit unions, financial non- governmental organizations (FNGOs), and cooperatives; 34 University of Ghana http://ugspace.ug.edu.gh 3. Informal suppliers of microfinance (e.g. susu collectors and clubs, rotating and accumulating savings and credit associations (ROSCAs and ASCAs), traders, money lenders and other individuals). In an attempt to sanitize the microfinance industry in the wake of recent Ponzi scheme associated with some microfinance institutions such DKM, the Bank of Ghana in March 2017 issued an Operating Rules and Guidelines for Co-Operative Credit Unions and Co- Operative Financial Institutions (CFI) in Bank of Ghana NOTICE NO. BG/GOV/SEC/2017/06. This notice contains guidelines on credit union management in relation to regulated activity, regulatory requirements, prudential oversight, submission and audited financial statement and prohibited actions. Interested readers can see appendix 1 for the details in the operating rules and guidelines notice for all credit unions. 2.3 Theoretical Models of Credit Unions In the theoretical examinations of credit unions, various authors have attempted to use the objective that the credit union seeks to achieve as the anchor on which to theorize or model the credit union. We identify five main views of the credit union discussed in the literature: the asset maximization view, profit maximization view, the cost minimization view or sales maximization view, the benefit maximization and benefit equality rate view and the benefit maximization and benefit equality value for each period view. We present the first four in brief; however we discuss the last one in some detail since it serves as the theoretical frame work for this thesis. 35 University of Ghana http://ugspace.ug.edu.gh The Asset maximization view supported by Smith (1971) and Chateau (1980) sees credit unions as being managed to maximize the asset of their owners. Although it is true that asset maximization does play an important role in the credit union management, it goes against the widely acclaimed objective which is at the center of the credit union definition as a social not-for-profit organization serving the financial interest of members. The profit maximizing view by Hempel and Yawitz (1971) holds that credit unions should be seen as enterprises that exist to maximize profit from their operations. This presentation of the credit union goes against the non-for-profit cooperative nature of credit unions. Cost minimizing or sales maximizing holds the assertion that credit unions exist to minimize the cost of mobilization funds, hence charging low loan rate and paying average deposit rate. The proponents of this view include Taylor (1971, 1977, and 1979), Koot (1976) Murray and White (1980) Murray and White (1983), Navratil (1981) and Fry et al. (1982). The benefit maximization and benefit equality for each period is based on the management objective of aligning the performance of the credit union by comparing best alternative loan rate less credit union loan rate, and credit union deposit rate less best alternative deposit rate. Flannery (1974) Walker and Chandler (1977) Kohers and Mullis (1987) are cited to support this proposition. Benefit equality here means both the 36 University of Ghana http://ugspace.ug.edu.gh borrower and the saver in the credit union share all the benefits equally in the form of low loan rates for borrowers and high deposit rate for savers. This situation hardly exists in banks and other depository financial institutions and some other mutually owned organizations. The benefit maximization and benefit equality for each period refers to best alternative loan rate less credit union loan rate times credit union loan volume, and credit union deposit rate less best alternative deposit rate times credit union deposit volume per Smith, Cargill, and Meyer (1981), Smith (1981,1984 and 1988) Gooptu and Lombra (1987) and Rubin et al, (2013). The difference here from the model proposed by Flannery (1974), Walker and Chandler (1977), and Kohers and Mullis (1987) is that the difference in rates on loans and deposits multiplied by the volume of loans and deposits respectively and concentrates on the values created thereof. These values matter to the owners of the credit union and not just the difference in the rates on competing loans and competing saving rates. 2.3.1 Static theoretical framework of the Credit Union Decision Making. In making a theoretical case for credit unions (CU), this study adopts the approach used by Smith (1984). In this work Smith established that, the credit union (CU) objective for the period is to choose the best loan rate rL to be charged the borrower compared to competing market loan rates rLM , offer the best deposit rate rS to savers compared to competing market deposit rate rSM . The loan rate charged and the deposit rate offered by 37 University of Ghana http://ugspace.ug.edu.gh the credit union is subject to the credit union’s retained earnings for the period and surplus or losses  incurred for the period. The objective function of the credit union according to Smith is given as, 1 d 1 d rL ,rS ,  rLM  rL L  1   rS  rSM S  (2.1) d  d  From equation 2.1 the left term of the equation, rL ,rS ,, presents the objective of the credit union, the loan rate rL charged and the deposit rate rS set by the management of the by the credit union, where λ is the Lagrange multiplier. The first term on the right of the equation, is the coefficient of the loan value extended during the period.  is used to capture the preference of the credit union towards lending where 0   1 . When  1 the credit union is totally oriented to the interest of its borrowers; a credit union that is totally oriented to the interest of its savers is represented by  0 . A neutral CU as described by Flannery (1974) places equal weights on gains to 1 borrowers and savers and is modelled as   . The present value factor for loans is 2  1 d    . Competing loan rates on the market is rLM , net gains on loans is given as  d   rLM  rL L , with L as the value of loan extend at the end of a given period. The second term on the right side of equation 2.1 is the coefficient of the savings mobilized during the period 1   is the preference of the credit union towards savings. 38 University of Ghana http://ugspace.ug.edu.gh  1 d  The present value factor for savings is   . Available competing savings rate on the  d   market is rSM and the value of savings mobilized at the end of a given period is S . The managers of the credit union can maximize the net gain of the credit union choosing loan rates rL and subject to the statement of financial position and rS dividend rates subject to the operating income statement constraint. Where the  is the profit equation which is a combination of the statement of financial position constraint and the income statement. The rationale in equation 2.2 is that the  , profit is the surplus captured after the cost of creating and managing loans and the cost of mobilization of savings, in addition to investment created and the general cost of running the credit union for a given period. The surplus is then transferred to the equity in the statement of financial position. This process is captured in equation 2.2 as   rL0 1   cL0  rIM 1 L0  rL 1   cL  rIM L  rSO  cS 0  rIM 1S0   rS  cS  rIM S  rIM   U0  rIM R0 C (2.2) The left side of equation 2.2 is  which is the combination of the statement of financial position constraint and the profit constraint. The first term on the right hand side of the profit equation , rL0 1  cL0  rIM 1L0 is net revenue from the inherited loan, rL0 is weighted average of past period CU loan rates,  is a fraction of gross revenue 39 University of Ghana http://ugspace.ug.edu.gh transferred to the regular reserve each period, cL0 is old loan amount and the average cost of managing old loan amount and rIM is investment rate for mostly liquid financial asset available on the market the that the management of the credit union can invest excess deposit over loan demand. The variable  is the proportion of loans inherited from the previous period that is retired during the current period. The second term rL 1 cL  rIM L is current issued loan portfolio, cL is new loan amount and the average cost of managing new loan amount. The interest on loans after any refund is rL and L is the value of loans extended in the period. The 3rd rSo  cSo  rIM 1So is the total cost of savings deposit, including those inherited from the previous period and 4th term rS  cS  rIM S is new savings deposit mobilized during the period. The dividend rates on previous period deposit and current period deposit are rS 0 and rS respectively. The variable  , is the proportion of savings balance inherited from the previous period that is retired during the current period, hence saving withdrawn at the start of the period S0 earn no dividends, 1 is the proportion of savings not withdrawn from the previous period, cS 0 and cS are the average cost of maintaining, promoting and issuing the savings account, rIM is the rate on liquid financial investment available on the market. The 5th term rIM U0 is the portion of undivided earnings that is invested from the previous period. The U 0 , is the undivided earnings which can be positive or negative. In 40 University of Ghana http://ugspace.ug.edu.gh this formulation the determination of  is not explicitly modeled. The 6th term rIM R0 is  the amount of reserves that is invested. The variable C is the fixed cost of operation and could include expenditure for nonfinancial member services such as investment counseling, estate planning, tax preparation and social activities. 2.3.2. Dynamic framework of the Credit Union Decision making. Building upon Smith and extending the credit union decision making beyond the one period case in order to capture the dynamic nature of the credit union objective, Rubin, Overstreet, Beling and Rajaratnam (2013) present a dynamic form of Smith’s work with some slight modifications. In their intertemporal model, the credit union chooses loan rate, lcu t deposit rate functions dcu t , and equity Et to maximize as follows  PE ,L t U BV S  e dt (2.3) 0 Equation 2.3, can be written as  U Llm  lcu V Dd  d  e PE ,Lt cu m dt 0  U LtltV Dtdt ePE ,Lt dt (2.4) 0 The first constraint is the statement of financial position constraint given as L I  DE (2.5). 41 University of Ghana http://ugspace.ug.edu.gh The statement of financial position identity is specified in equation 2.5, the left side of the equation is the asset of the credit union made of L is loan volume, I is investment, the right side of the equation is the liabilities composed of D is deposit volume and E is equity. The second constraint is the equity constraint given as dE  lcu L  rI  dcu D CL, D (2.6) dt The equation in 2.6 is the net surplus of the credit union at the end of a period. The left dE side of the equation , is the change in equity position of the credit union over time, dt on the right side of the equation, lcu L is the interest income on loans, rI is the investment income, the second term dcu D is the cost of interest on deposit paid on savings. The variable CL, D is the fixed cost associated with making loans and mobilization deposit and  is the tax rate applicable at the end of the period. From (2.3), B  Llm  lcu  is loan benefit to the borrower from the credit union. Similarly, S  Ddcu dm  is saver benefit from maintaining a savings account at the credit union. Also let lt lm  lcu  and dt dcu  dm  . The discount rate for a typical   credit union is p . The U  is borrower utility function, U  0 and U  0 and saver   utility function is V V  0 and V  0 42 University of Ghana http://ugspace.ug.edu.gh What is clear from these theoretical postulations of the credit union decision making from the static one period Smith (1984) and Ruben et al, (2013) dynamic credit union decision making process is that loan generation, savings mobilization by the management of the credit union recognizes the competitive market place for the credit union. From this credit union business is conducted with the objective of maximizing the benefits that accrue to the borrower from the credit union and maximizing the benefits that accrues to savers from higher deposit rates whiles being aware of the activities of competing financial institutions. The credit union resource usage and constraint is also captured in these optimization models. Readers interested in the details are referred to the works of Smith 1984 and Ruben et al (2013) for the detail analysis of the credit union decision making. 2.4 Theoretical Review on Income Diversification Henry Markowitz’s 1952 Modern Portfolio Theory (MPT) provided the concept of financial diversification with its theoretical foundations. According to this theory, diversification, if done properly, should reduce the overall risk of an investment portfolio and stabilize returns. The diversification principle demands that in building an investment portfolio asset with different risk returns, those characterised by mostly negative correlation should be combined with positively correlated asset to form portfolios which then reduces the overall risk and provides a stable return on the portfolio over a given period of time. Pitts and Hopkins (1982) characterise diversification as the extent to which firms operate in different businesses simultaneously. Booz, Allen and Hamilton (1985) provide a broad 43 University of Ghana http://ugspace.ug.edu.gh definition of diversification which includes the goals of diversification; in their viewpoint, diversification is a means of spreading the base of a business to achieve improved growth or reduce overall risk. The latter includes all investment except those aimed directly at supporting the competitive position of existing business, investments that address new products, services, customers or geographic markets and which may be accomplished by different methods including internal development, acquisition, joint venture, and licensing agreements, among others. The diversification literature is vast, covering different aspects of the financial institution industry. This study briefly explain three theories of diversification viz, market power theory, agency view and resource theory, other theories include the financial view and synergistic view of diversification. In the Market power theory of diversification, companies that have large amount resource are naturally positioned to make use of these resources to control the market; these companies diversify their activities and in some instances engage in various forms of anti-competitive behavior. According to Montgomery (1994) the sources of this market power include, cross subsidization, mutual forbearance and reciprocal buying. The market power view of diversification does not fit the credit union set up. The Agency view holds that, it is the interest of managers (agents) with no considerable ownership stake who engage in growing the enterprises by venturing into other activities with the view of enhancing shareholder value through diversification. In the end 44 University of Ghana http://ugspace.ug.edu.gh shareholders do not benefit because these managers engage value reducing project (Mueller, 1969) and empire building which makes them powerful to the extent that the enterprise end up depending on them to manage these large organisation with more resource been allocated by shareholders to pay managers for managing these enlarged groups. The counter view is that shareholders should be paid cash dividend so they can diversify their own portfolio which is much cheaper than managers diversifying business activities at the corporate level. The reasons for managers undertaking diversification included empire building, managerial entrenchment, employment risk reduction and the use of free cash flow. With the managers of credit unions mostly members and the democratic nature of credit unions, the agency view cannot be ascribed for the credit union. The Resource view relates to the resources endowed with a firm not the market or industry the firm operates in. In Goddard McKillop and Wilson (2008) resources refers to the specific asset, core competences or distinctive capabilities of the firm that can, potentially, be exploited in new markets to rake in greater profits hence diversification. We hold the view that the resource theory can fit the credit union attempts at diversifying their income stream. Of special focus has been banks’ operational performance and much of the empirical evidence in the literature relates to the US and the European banking industry although a few studies exist on banks in emerging markets. This review is guided by the study’s 45 University of Ghana http://ugspace.ug.edu.gh objective attempts to cover how diversification relates to financial performance, agency and resource usage, risk, size and financial stability. 2.4.1 Revenue Diversification and Financial performance. Diversification and financial performance have been amply discussed in the literature. For most of these studies diversification is measured by non-financial income while financial performance is measured on a return basis, namely return on asset (ROA) or return on equity (ROE), and on a risk adjusted basis, namely risk adjusted return on asset (RAROA) and risk adjusted return on equity (RAROE). The available evidence has proffered varying results, some authors presenting a positive relation between income diversification and performance while others counter claim that diversification results in decreased performance. For developed economies, studies such Chiorazzo et al (2008) analyzed evidence from Italy and demonstrated that between 1993 and 2003, Italian banks exhibited an increase in profitability measured by risk adjusted performance when the banks diversified their income generating activities, with non-interest income contributing largely to bank performance. Busch and Kick (2009) argue that German Universal and Commercial banks that engaged in high fee income activities also reported an increase in risk adjusted return on equity and total asset, notwithstanding the fact that higher fee-income activities generated a much higher risk for these banks. Similarly, DeYoung and Rice (2004) provide evidence from studies on US banks that incremental increase in non-traditional bank income were linked with higher performance measured by bank profit. 46 University of Ghana http://ugspace.ug.edu.gh For emerging markets, Senya and Wolfe (2011) advance the argument that banks which diversify their income sources recorded some positive gains in performance. This view is further strengthened by Meslier, Tacneng and Tarazi (2014) who examine the benefit of bank income diversification in an emerging market context and conclude that a shift toward non-interest activities increases bank profits and risk-adjusted profits particularly when banks are more involved in trading in government securities. If the level of financial development is an issue to consider, then Doumpos, Gaganis and Pasiouras (2016) adequately extend the evidence to cover bank diversification and overall financial strength. Their study reveals that diversification in terms of income, earning assets and on and off-balance sheet activities positively influence bank financial strength. Additionally, income diversification can be more beneficial for banks operating in less developed countries compared to banks in advanced and major advanced economies. Brighi and Venturelli’s 2014 investigation into how a mix of various non-traditional revenues impact performance found that diversification increased bank profitability on a risk-adjusted basis; however, they did not observe any statistical effect in terms of risk. Edirisuriya, Gunasekarage and Dempsey (2015) on specific Australian bank features, income diversification and performance, stressed that there is no strong evidence to suggest that diversification has been unfavourable to the performance of Australian banks; rather, Australia's banks have improved their risk-return profiles as an outcome of income diversification. 47 University of Ghana http://ugspace.ug.edu.gh To what extent should banks diversify into non-interest income to enjoy the benefit? In their attempt to answer this question, Gambacorta, Scatigna and Yang (2014) analyze the nonlinear relationship between non-interest income to total income and bank return on asset and conclude that income diversification is positively correlated with bank profitability only up to about 30 percent of bank revenue diversification ratio. Nguyen, Nguyen, Nghiem and Nghiem (2016) employed a two - stage approach with double bootstrap method to examine the operational efficiency and effects of market concentration and diversification on the efficiency of Chinese and Indian banks from 1997 to 2011. Their findings confirmed that in the case of Chinese banks, diversification of revenue, earning asset and non-lending earning assets are associated with increasing profit efficiency, but their effects to cost efficiency are not clear. The signal on Indian banks is that diversification of earning assets increases profit efficiency while there are cost efficiency losses from diversification of revenue and earning assets. From the foregoing empirical evidence, banks in developed economies and emerging economies benefit when they diversify their income, an outcome mainly observed in increased performance. For the credit union milieu, Goddard, McKillop and Wilson (2008) measured performance on both risk-adjusted and unadjusted returns and reported a positive direct exposure effect being higher than a negative indirect exposure effect for all but the largest credit unions. Thus they infer that similar diversification strategies cannot be suitable for large and small credit unions therefore small credit unions should eschew diversification 48 University of Ghana http://ugspace.ug.edu.gh and continue to operate as simple savings and loan institutions, while large credit unions should be encouraged to exploit new product opportunities around their core expertise. Mathaya (2016), in a study on revenue diversification and financial performance that included credit union in the unit of analysis, proffer evidence that in Kenya increased dependence on non-interest income is positively related to higher returns. In the US, Malikov, Zhao and Kumbhakar (2017) point out that the 27 to 91 percent of credit unions that have diversified exhibit the benefit of high levels of economies of diversification. They add that there exists a presence of non-negligible economies of diversification of financial services in the credit union industry. Another strand of these literatures is authors who hold the view that diversification lead to a reduction in financial performance of financial institutions. Stiroh (2004a) opines that return increase in fee based revenue leads to worsening risk – return trade off results. Additionally, Deyoung and Roland (2001), De joughe (2010) and Florderlisi et al (2011) have all presented evidence that reveal that banks experience a reduction in performance when they diversify income sources from traditional into non-traditional income generating activities. From their study of Italian banks, Acharya et al (2002) clearly show that banks cannot use diversification of asset to secure superior performance or to reduce risk. Esho et al’s 2005 study of Australian credit unions adds to the existing evidence that increases in revenue from share of transaction fees increases risk and reduces returns, while those that increase residential lending revenues reduce both risk and returns. Other 49 University of Ghana http://ugspace.ug.edu.gh studies that bolster the argument that diversification reduces financial performance include Stiroh and Rumble (2006) and the Reserve Bank of Australia (2005). Similarly, Mercieca et al (2007) found no positive gains from income diversification with regard to bank performance as did Trujillon-ponce (2013) whose assessment found no link between diversification and financial performance. 2.4.2 Diversification and Agency The empirical literature on diversification and agency follows the theoretical view that management benefits when they diversify the activities of their organizations. The facts, as available in the literature, clearly show that most income diversification activities do not benefit owners of financial institutions. Income diversification is expected to reduce variability in earnings if the income source is not perfectly correlated. The benefits in most cases go to the managers who choose to maximize their personal utility instead of shareholders’ wealth maximization (Berger et al, 1999; Bliss and Rosen, 2001; Aggar and Sawick, 2003). Aggar and Samwick (2003) provide the insight that banks that engage in merger activities increase owners risk to non – traditional income generating activities. In some cases diversification is seen as a decreasing value activity undertaken by management in an attempt to decrease their personal risk (Deng and Elyasianni, 2008; Laeven & Levine, 2007). The value reduction activity undertaken under the guise of diversification of income sources can be tracked all the way through increasing agency cost as shareholders must spend more resources to monitor the additional income generating activities undertaken by management. This selfish position of managers can 50 University of Ghana http://ugspace.ug.edu.gh also be attributed to the organizations’ ineffective incentive and reward schemes which could then provide the impetus for top managers to maximize their interests ahead of shareholders’ wealth maximization (Beger et al, 2010). Furthermore, additional income generating activities may demand knowledge and expertise that management may not possess, nullifying any expected benefit from diversification and potentially wasting the organisation’s resources. To manage this increase in agency cost, Jensen (1986), Beger and Ofek (1996) and Denis et al (1997) instruct managers to use their expertise to focus on single line products and specialize in their income generating activities. Eschewing management’s non-expertise, Acharya et al (2006), on the other hand, argue that the projected value from economies of scope are not achieved due to ineffective monitoring and non-performing loans when risky banks endeavor to new industry and business. Owners of financial conglomerates end up paying too high a price for engaging in multiple lending activities. This premium paid tends to lower the market value of these financial institutions, a situation which has compelled Leaven and Levine (2007) to propose that it would be prudent if these financial conglomerates were broken into and managed as separate financial institutions. The support for this assessment is that the economies of scope expected hardly materialized in the end, making it difficult to justify 51 University of Ghana http://ugspace.ug.edu.gh the premium paid by the shareholders, and in the case of improper due diligence, the agency problem is manifested. Trinarningsih, Husa, Untoro, Trinugroho and Sutaryo (2016) researching how top management teams affect diversification and bank performance, advanced the argument that diversification is negatively associated with performance; however, they find little evidence to support the moderating effect of top management team characteristics. Saghi- Zedek (2016) examined whether the presence of some categories of controlling shareholders affects product diversification performance and finds that when banks have no controlling shareholder, or have only family and state shareholders, activity diversification yields diseconomies. On the other hand, as long as the control chain involves banking institutions, institutional investors, industrial companies or any other combination of these shareholder categories, banks benefit from diversification economies: they display higher profitability, lower earnings volatility and lower default risk. The reason being that banks owned by varied shareholders supply supplementary skills and knowledge to manage new and diversified activities that the financial institution is engaged in. In the author’s opinion, the case of credit unions is akin to a “natural strong and natural weak” condition, as ownership is concentrated among members based on a common employer, community or societal bond. The pool of owners restricts management quality to the bestowed individual member’s quality as the credit union is mandated to appoint 52 University of Ghana http://ugspace.ug.edu.gh management from the pool of owners. From this, good member quality by way of knowledge, experience and expertise may go a long way to influence resource deployment; hence the credit union loan and income drive. A credit union with a pool of members embedded with weak member quality by way of too little knowledge, investment inexperience and lack of special expertise in the same vein may impact the credit union performance negatively despite regulatory provisions and protection. 2.4.3 Diversification and Risk From the theoretical point of view, diversification of asset is expected to reduce the overall risk of a portfolio. The empirical literatures that exist on income diversification by financial institutions, especially banks, mostly point to the same conclusion although other evidence exist that suggest a positive relationship between diversification of income and income volatility. The first group of empirical literature that suggests that bank income diversification reduces risk includes: Rose (1989) who claims that banks moving into non-bank product line could reduce cash low risk, a possibility indirectly supported by Templeton and Serveriencse (1992); Froot and Stein (1998) who argue that diversification is a hedge against insolvency risk that reduces the profitability of costly financial distress for banks that engage in nontraditional income sources and Williams (2016) who studied the relationship between bank revenue composition and bank risk in Australia found that those banks with lower levels of non-interest income and higher revenue concentration have less risk. 53 University of Ghana http://ugspace.ug.edu.gh The type of nontraditional income activities can also influence the nature of relationship when it comes to risk implication. In DeYoung and Roland’s 2001 study of large banks in the US, shifts in revenue diversification into the area of fee-based activities was linked with an increase in volatility and high leverage resulting in greater earnings volatility and concluded that there is no benefit from diversification. Looking at nontraditional income activity in particular, Sunders and Walters (1994) found that expansion of banks activities reduced risk and that diversification benefit from banks indulging in insurance activities reduces risk more than benefits from activities in the securities market. This particular conclusion is also reaffirmed by Casu, Dontis‐Charitos, Staikouras and Williams (2016) who examined the relationship between risk and diversification through determinants of risk bank acquisition and mergers insurance companies and securities firms from 1991 to 2012. They averred that bank combinations with securities firms yield higher risks than combinations with insurance companies. Rosen et al (1989) make the case that although banks’ risk increase as banks involve themselves in real estate activities, banks that heavily engage in real estate expansion tend to earn more. Further evidence for other types of non-interest income activity is found in Sinkey and Nash (1998) who examined fees generated from credit card lending. They argue that commercial banks in this area of additional revenue have more volatile returns albeit with higher profitability of insolvency compared to banks that focus on traditional product lines for income. DeYoung and Rice (2004) adds that non-interest 54 University of Ghana http://ugspace.ug.edu.gh income increase bank profit variability and has a worsening effect on bank risk return trade off. Stiroh and Rumble (2006) make a worsening risk return trade-off case for US banks as gains from non-interest income are way below the volatility of earning experience, which in the end makes the stock return of these listed banks low considering the level of exposure that these banks have. This conclusion is similar to Stiroh (2004) that banks in the US that take solace in non-interest income in particular reference to trading income, leads to higher risk and with less diversification benefit. Batten and Vo (2016) show that those commercial banks that have shifted to non-interest income activities face higher levels of risk. These set of findings is at variance with theories and evidence that argue that diversification is a strategy for risk reduction. In most of these empirical studies, banks reduce their unique or diversifiable risk because managers are able to move into imperfectly correlated income generating activities. While this helps, it also exposes and increases the systemic or market risk of the bank. De Vires (2005) makes this clear by showing that there is a corresponding rise in exposure of systemic shocks because of the number of markets banks become active in. Other authors including Dermrguc-kunt and Huizinga (2010) and Baele et al’s (2007) studies of banks in Europe point to the fact that higher level of non-interest income for banks have resulted in an increase in risk for these banks. In the study on 62 main Chinese banks over the period 1997 to 2012 banks, Zhou (2014) shows that that there is no significant relationship between income diversification and bank risk. 55 University of Ghana http://ugspace.ug.edu.gh The issue of bank size and diversification has also been looked at in the literature. According to Demestz and Strahan (1997), large banks are more diversified compared to smaller banks. Chiorazzo et al (2008) add that diversification gains decline with banks size, inferring that gains are stronger for larger banks than for smaller banks but also records that small banks with very small non-interest income shares gain the most from diversification. However, this conclusion has been countered by Mercierca et al (2008) who posit that small banks do not benefit from diversification. Earnings stability is also a key reason why managers of deposit taking financial institutions would diversify income sources and assets. Smith et al (2003), examining banks in Europe for the period 1994 to 1998, found that when banks effectively combine interest income and non-interest income activities, non-interest income potentially stabilize bank earnings. DeYoung and Torna (2013) add that the shift towards non-interest income compromises bank stability and exposes banks to high probability of failure, especially during period of financial crises for banks that have a weak financial position. 2.5 Theoretical Review on Lending The literature on lending presents two classes of information used by lending financial institutions and banks in particular when granting and monitoring loan extended to borrowers. Ranjan (1992) asserts that lenders sometimes rely on soft information, that is, qualitative information obtained by means of personal interactions to evaluate a client’s ability to repay loans. On the other hand, Berger and Udell (2006) emphasize hard 56 University of Ghana http://ugspace.ug.edu.gh quantitative information, such as information derived from the borrowers’ statement of financial position, for assessing a borrower’s loan servicing ability. The theme of lending in the corporate finance and banking literature is broad; however, for the purposes of its objectives, this study concentrates on lending as presented in the banking literature, viz loans granted to borrowers. The study focuses on two lending propositions, namely transaction based lending and relationship lending. 2.5.1 Transaction Based Lending According to Bolton, Freixas, Gambacorta and Mistrulli (2016), in transaction banking, banks specify a gross repayment at the inception of a loan. If the borrower defaults, the lender has the right to liquidate the firm’s assets to repay the loan contracted. At the repayment date if the borrower defaults, the bank can also offer to roll over the defaulted loan to the next repayment period. The sum to be paid at this point is the amount that the borrower’s expected cash flow can comfortably accommodate if the loan is not to fail. The bank in this situation has the right at any time to liquidate a borrower’s assets to pay for defaulted loans. Transaction banking mostly takes in competitive banking market setting as no bank has information advantage over other competing banks in the provision of financial services and products. In sum, transaction-based banking treats each loan deal as a single contract and focuses on the risk of the loan. 57 University of Ghana http://ugspace.ug.edu.gh The capital allocation model by Stein (2002) posits that large complex banking organizations will not engage in relationship lending because these banks cannot verify soft information produced by the borrower and considers the cost of verifying such information as too expensive to be incurred on a small business client. Berger, Miller, Petersen, Rajan and Stein (2005) provide evidence that small banks are better positioned to collect and act on soft information than large banks as large banks are less willing to lend to informationally ‘‘difficult’’ credits, such as firms with no financial records. They further contend that large banks lend at a greater distance, interact more impersonally with their borrowers, have shorter and less exclusive relationships, and do not alleviate credit constraints as effectively, especially for the small business. From the foregoing discussion, small business credit needs can be said to be the preserve of small sized financial institutions as large size banks would prefer doing business with well-established clients, a strategy that reduces information asymmetry problems and tackles moral hazard challenges in the lending business. The credit union, being concentrated by way of ownership, clientele base and localized depository financial institution, naturally has an incentive to provided credit facility to owners of small size business who are members of the credit union. The credit union’s ownership-clientele nexus makes its lending business akin to the relationship lending type than the transaction lending one. 58 University of Ghana http://ugspace.ug.edu.gh 2.5.2. Relationship Lending Financial institution lending can be related to relational contracts which have the aim of building long-term relationships between the borrower and the financial institution, to establish strong business interest for both parties. Under this concept, the bank charges high loan rate at the inception of the lending relationship but lowers this over time as the borrower repays loans. In the interim the bank gathers more information on the borrower reducing the information asymmetry and, subsequently, the loan rates. With the improvement of information banks collect progressively, loan rates are reduced for both new and existing firms (see Bolton & Sheristein 1996 and Boot & Tharkor 1994). Embedded in relational contract is relationship banking. According to Berger and Under (2002), banks acquire information in a relationship lending setting through contact with the firm, its owner, and its local community on a variety of dimensions and use this information in their decisions about the availability and terms of credit to the firm. This customer interaction can be through the provision of financial products, deposits made by customers and contacts with suppliers and customers who give specific information about the firm. In relationship banking there is 'social attachment' in conducting the lending business with the borrower as the loan officer plays a central role in the lending process. The loan officer assesses the borrower’s banking relationship and other social commitments of the borrower to build a good picture of the borrower’s ability to service prospective loan facilities. Sharpe (1990) opines that during the later stages of relationship lending, 59 University of Ghana http://ugspace.ug.edu.gh lending institutions, having built a substantial enough dossier on borrowers, tend to exploit this information to borrowers’ disadvantage. 2.5.3 Macroeconomic Environment In macroeconomic considerations of financial institutions’ lending, various hypothesis exist that assess how monetary policy impacts on bank lending. The interest rate channel proposition argues that a tight monetary policy will lead to an increase in short term nominal interest rate which would in the short term lead to an increase in real interest rate. The conventional interest-rate channel stresses the direct impact of interest rates on loan demand. Per Gambacorta (2005), monetary tightening leads to larger reductions in loan supply for small banks. The bank lending channel holds the view that, when banks are presented with a funding shock in the case of monetary tightening, banks may fall on liquid funds to make up for the short fall in loanable funds; if not, the supply of bank loans may be reduced. Subscribing to the bank lending channel, Bernanke and Blinder (1988) contend that monetary policy actions affect the balance sheet structure of banks, causing changes in banks’ loan supply in addition to causing changes in loan demand. In situations of reduced loan supply by banks, small and medium scale businesses tend to suffer the most (Kashyap & Stein 1994). Bernanke and Gertler (1995) premise their work on the balance sheet channel which suggests that businesses with lower levels of equity would aggressively pursue a bank for 60 University of Ghana http://ugspace.ug.edu.gh loans, heightening the adverse selection and moral hazard phenomenon. Aware of this, banks cut back lending to low equity finance firms. On the other hand, if the equity stake in the business firm is high, then the adverse selection and moral hazard problem is reduced and the bank may be prepared to increase its lending to this firm. Pruteanu‐Podpiera (2007) observes that changes in monetary policy in the Czech Republic altered the growth rate of loans, with considerably stronger effect in the period 1999–2001 than in the period 1996–1998. Furthermore, cross-sectional differences that exist in the lending reactions to monetary policy shocks could be attributed to the degree of bank capitalization and to liquidity. Italian financial intermediaries, according to Quagliariello (2007), expand their lending activity which invariably affects their loan loss provision and new bad debts as a result of business cycle evolution. From the preceding it is clear that monetary policy does influence financial institutions’ lending decisions. 2.5.4 Other Factors In terms of industry effect, Cecchetti (1999) opines that the strength and the degree of the supply of bank loans also can depend on the size and the concentration of the banking system. Smirlock (1985) finds a positive relationship between size and bank supply of loan, implying that banks with larger asset size give out more loans compared to small sized banks. On the other hand, Vihriälä (1997) finds that the lower the degree of capitalisation of a bank, the more expansionary the supply of loans. For Ehrmann, Gambacorta, 61 University of Ghana http://ugspace.ug.edu.gh Martinez‐Pagés, Sevestre & Worms (2003) there is no association between bank size of European banks and supply of loan and also offer the view that competition is the conduit for monetary policy to affect bank lending. They also claim that the transmission of monetary policy is through the bank lending channel which is less pronounced for banks with extensive market power. Petersen and Rajan (1995) do not make a compelling case for competition on bank relationship lending. Faleye and Krishnan (2017) study the effect of bank governance on risk-taking in commercial lending and find that banks with more effective boards are less likely to lend to risky borrowers. According to Amidu (2014), the structure of banking markets influences credit delivery in Sub-Saharan Africa in an environment where the financial sector is reformed and banks are allowed to operate freely and suggests a link between bank credit and the financial strength of the banks. The overall results from Amidu suggest that regulatory initiative restricts banking activities, imposes severe entry requirements and requires high regulatory capital, influences banks’ decisions to supply loans. 2.5.5 Credit Union Lending From a theoretical standpoint, the nature of credit unions, in that they transact business with owners, can render credit union lending closely akin to relationship lending rather than transaction based lending. However, the literature on credit union lending is scanty and examines other aspects of credit unions’ lending practices. For example, Sayles (2002, cited in Wilcox and Berkely, 2011), employing a case study methodology, 62 University of Ghana http://ugspace.ug.edu.gh examines the concentration of business loan programs in credit unions. Ely and Robinson’s 2009 work attempts to allay small business managers’ fears that the extension of credit would decrease sharply in the face of the takeover or consolidation of community banks by larger banks. Addressing the concern that these small businesses would be overlooked by large banks, Ely and Robinson claimed that the ability of small businesses to access credit may be dependent on the remaining unmerged community banks, including credit unions, and their decision to extend credit facility to small businesses. They concluded that credit unions would probably provide loans to small businesses in markets dominated by greater bank consolidation, implying that the credit needs of small business would be left unattended to in these markets. In an effort to model the credit union loan stance, Emmons and Schmid (2000) employed a dynamic model of spatial competition for bank and credit union in a for-profit making and a not- for-profit making setting to analyze competition among banks and credit unions. Providing empirical evidence of two-way competitive interactions between banks and credit unions, their study showed that credit union participation rates are higher in more concentrated local deposit markets. In the Emmons and Schmid (2000) study, high bank concentration measured by the Herfindalhl index, highlights the fact that higher prices for financial services offered by banks influences consumers to move from banks to credit unions for financial service, thus negatively impacting bank service provision. According to Emmons and Schmid, this in turn unleashes a continuing endogenous adjustment process of higher bank concentration and increasing credit-union 63 University of Ghana http://ugspace.ug.edu.gh participation. In their opinion, credit unions are straight forward competitors to banks in the retail financial services market. As a result of losing market share to credit unions, banks respond by committing more resources in local deposit market which, Emmons and Schmid hypothesize, leads to higher credit-union participation and this increase in credit-union participation leads to higher deposit-market concentration. The competitive position of credit unions in the provision of financial services is corroborated in Feinberg (2001) who argued that as the number of credit unions in a particular market, increase rates charged by banks on new vehicle consumer loans fell, suggesting that credit unions play a significant role in disciplining the exercise of market power by banks. This conclusion was drawn from a pooled cross-sectional, times-series study on 1,000 observations on relatively small United States markets. Using binomial logistic analysis, Adusei and Appiah, (2011) analyzed cross-sectional data on 222 credit unions in Ghana for the year 2008 and find that management size, lower repayment performance, no delinquent loans over 30 days, better liquidity positions, and longevity of credit unions influences the likely adoption of group lending. 2.6.1 Theoretical review Cost Efficiency and Technical Efficiency The concept of efficiency in the economics and finance literature is broad. This review is not meant to be a comprehensive review of all the types of efficiency; rather, based on the objectives of this study we concentrate on Cost Efficiency (CE) synonymous to Economic efficiency and Technical Efficiency (TE). 64 University of Ghana http://ugspace.ug.edu.gh From the work of Banker, Charnes and Cooper (1984) efficiency defined in a multiple- input, multiple-output framework, involves obtaining the maximum outputs from a given set of inputs, or minimizing inputs required to produce current output levels. Also Maudos, Pastor, Perez, and Quesada (2002), refers to cost efficiency as the ratio between the minimum cost at which it is possible to attain a given volume of production and the realized cost. A decision making unit (DMU) is cost efficient if it uses minimum level of cost as compared to the actual cost observed at the end of a production period. We can in this case measure cost efficiency by using the ratio between the minimum level of cost of the potentially efficient credit union and the cost level actually observed. CE which can be decomposed into both Technical Efficiency (TE) and Allocative Efficiencies. TE captures managerial efficiency and can further be divided into Pure Technical Efficiency (PTE) and Scale Efficiency (SE) under variable returns to scale (VRS). AE or Price Efficiency (PE) captures a firm’s ability to use inputs in optimal proportions or mix of inputs given their respective prices. Koopmans (1951) provides a formal definition of technical efficiency as an increase in any output that requires a decrease in at least one other output or an increase in at least one input, and if a reduction in any input requires an increase in at least one other input or a reduction in at least one output. Offering a different perspective, Debreu (1951) and Farrell (1957) provide the following measure of technical efficiency, known as the Debreu-Farrell measure as “the maximum possible reduction in inputs when the output is 65 University of Ghana http://ugspace.ug.edu.gh given”. A technically inefficient producer then uses the same inputs to produce more of at least one output. A DMU is technically efficient only when it uses the least amount of inputs to produce maximum outputs. Or, it may use reduction in input levels while giving up the same amount of output. The method for estimating efficiency can be divided into two major groups: the econometric (parametric) and the mathematical programming (nonparametric) approach, the distinction between the parametric methods and non-parametric methods being that the former assign a density function to the stochastic component of the model, while the latter only define the deterministic part. The parametric approach includes Stochastic Frontier Analysis, Thick Frontier Approach, and Distribution Free Approach. The Stochastic Frontier Analysis (SFA) is a stochastic method that assigns a distribution to the error term. In this approach, the distance of each DMU from the frontier is considered inefficiency and also ascribed to a random error according to Kumbhakar and Lovell (2000). To detach these two components, an asymmetrical probability distribution is assumed for the inefficiency term. Another important feature of SFA is that it allows the implementation of significance tests on the estimated parameters by assigning a distribution error. In empirical studies, the SFA appears preferable when compared with other parametric methods such as Thick Frontier Approach (TFA) because it estimates the inefficiency of any DMU and, therefore, gives an indication of the level of efficiency for the sector 66 University of Ghana http://ugspace.ug.edu.gh under analysis too. A further advantage of the SFA method is the ability to insert a set of variables into the model that explains the inefficient component. The Thick Frontier Approach by Berger and Humphrey (1991) does not make any distributional assumptions for the random error and inefficiency terms but assumes that inefficiencies differ between the highest and lowest quartile firms. The distribution-free approach (DFA) by Berger (1993) makes fewer specific assumptions but relies on extensive data set; efficiency of each DMU is assumed to be steady over time, and the random errors evens out to zero. The nonparametric technique includes Data Envelopment Analysis (DEA) and Free Disposal Hull (FDH) method. In the nonparametric approach less emphasis is placed on the structure and the specification of the efficient frontier and there is no decomposition of inefficiency and the error terms. The Free Disposal Hull (FDH) approach is a special configuration of DEA. Under this approach, the points on the lines connecting the DEA vertices are excluded from the frontier and the convexity assumption on the efficient frontier is relaxed from Berger and Humphrey (1997). In Lim, Lee and Lee (2016) the computational technique to solve FDH programme considers the mixed integer programming problem compared to the DEA model with a linear programming problem. DEA popularized by Charnes, Cooper and Rhodes (1978) from the frontier estimation seminal work of Farell (1957) is a mathematical linear programming based technique for 67 University of Ghana http://ugspace.ug.edu.gh measuring relative efficiency of DMU that have multiple inputs and outputs. It was originally developed for measuring the efficiency of not-for-profit, public sector enterprises but over time has found wide spread applications in the context of profit oriented organizations, especially the financial service industries including banks and insurance among others. In performance benchmarking, DEA is also used to identify the efficient DMU on the “best-practice frontier” compared to the “production frontier” in the case of production. DEA model can either be input or output oriented. In the output oriented DEA model a DMU is assumed to maximizing outputs while keeping inputs constant. In the input oriented DEA model, the DMU is assumed to minimizing inputs for a given amount outputs. The efficiency score in an input oriented DEA is above 0 but and not more than 1. DEA can be estimated with either the constant returns to scale (CRS) or variable returns to scale (VRS) assumption. In the non parametric DEA, there are two measures of technical efficiency with different characteristics: radial and non-radial. The radial measure is represented by the original Charnes, Cooper and Rhodes (1978) model in which efficiency is, Weighted sum of Outputs Technical Efficiency  (2.9) Weighted sum of Inputs The weights for the ratio are determined by the restriction that the similar ratio for every DMU have to be less than or equal to unity, thus reducing multiple inputs and outputs to a single “virtual” input and single “virtual” output without requiring pre-assigned 68 University of Ghana http://ugspace.ug.edu.gh weights. In employing the DEA technique, the researcher may seek to, among other things, compare performance of homogeneous decision making units that use multiple inputs for the production of multiple outputs, and estimate efficiency measure that compares the ratio output to input of the DMU assessed with the value of this ratio observed in the other DMUs analysed. DEA has several advantages: it is easy to use, it allows for multiple inputs and multiple outputs in a single framework, it does not require specification of functional form for the frontier, it does not require a priori specification of weights for inputs and outputs and, finally, inputs and outputs can be expressed in different measurement units. One weakness of DEA is the assumption that the distance from the frontier is due entirely to inefficiency without considering random errors such as errors of measurement or unforeseen events impacting on the DMU, including the fact that the technique is sensitive to outlier data. Further DEA does not measure "absolute" efficiency but relative efficiency and finally statistical hypothesis tests are difficult to conduct in the DEA method. 2.6.2 Efficiency Empirical Review Empirical literature on efficiency is widely spread across financial institutions using parametric and non-parametric estimations and further on, the use of regression technique to assess factors that influence efficiency. What follows is a review of the empirical literature on cost and technical efficiency relating to credit unions. 69 University of Ghana http://ugspace.ug.edu.gh Worthington (1998) uses limited dependent variable regression techniques to relate credit union efficiency scores to structural and institutional considerations in Australia. The results indicate that non-core commercial activities are not a significant influence on the level of cost inefficiency although asset size, capital adequacy regulation, branch and agency networks are significantly associated with cost efficiency. In a study using a sample of 233 credit unions, Worthington (1999) used non-parametric techniques to measure efficiency and proceeded to use parametric techniques to attribute variation in efficiency. The results indicated that a large number of credit unions in Australia were best-practice efficient, and any efficiency found appears to flow from x- inefficiencies rather than from the selection of an inappropriate scale of operations. In addition, the study predicts that credit unions formed on the basis of a community bond, with a large asset base and an orientation towards commercial loans will be relatively more efficient. In factors that impact efficiency, Worthington (2000) employed a two-stage procedure to evaluate non-bank financial institution cost efficiency and reveal that the major source of overall cost inefficiency appears to be allocative inefficiency rather than technical inefficiency. From the regression results, commercial lending activities, expenditures on information technology and marketing and promotion, the proportion of non-interest income, and association membership are a significant influence on the level of cost efficiency. 70 University of Ghana http://ugspace.ug.edu.gh Using similar credit union type, McKillop, Glass and Ferguson (2002) examined relative efficiency of UK credit unions using radial and non-radial efficiency measures to investigate cost performance. Their research showed that credit unions have considerable scope for efficiency gains using both measures of efficiency. Further, they opined that high level of inefficiency may be indicative of the fact that credit unions, based on clearly defined and non-overlapping common bonds, are not in competition with each other for market share. Esho’s (2001) sample of Australian credit unions demonstrated that bond type, size, age, average deposit size and interest rate spreads are significant determinants of relative cost efficiency after estimating efficiency using stochastic frontier and distribution free approach, noting that subsidies bias estimates of cost efficiency and efficiency rankings. Frame, Karels, and McClatchey’s work (2003) probed how credit unions use tax advantages in providing benefits to members. They estimate a translog cost function for credit unions and mutual thrifts focusing on the unique objectives of mutually owned depository institutions and show that credit unions with residential common bonds have higher costs than mutual thrifts; however they note that single common bond occupational and associational credit unions are more cost efficient. They observe that residential credit unions engage in expense preference behavior and hence redirect some portion of their tax benefits away from members. 71 University of Ghana http://ugspace.ug.edu.gh Using stochastic frontier analysis, McKillop, Glass, and Ward (2005) concluded that UK credit unions from the period 1999 to 2001 were subject to high levels of inefficiency and that the environment in which a credit union operates plays a vital role in relative efficiency. Furthermore, larger credit unions that operate in well endowed environments and membership were more cost efficient. Battaglia, Farina, Fiordelisi and Ricci (2010) analysed the impact of environmental factors on cost and profit efficiencies of cooperative banks including credit unions in Italy for the period 2000 to 2005. They estimate cost and profit efficiency using stochastic frontier analysis (SFA) and with various environmental variables accounting for disparities among Italian regions and show that environmental conditions substantially influence efficiency estimates. Cooperative banks in the northeast of Italy are shown to be the more cost efficient, benefiting from a favourable environment, while cooperative banks in the south of Italy display a higher profit efficiency, probably due to lower competitive pressures albeit branch network and concentration of the industry impacting both cost and profit efficiency. Wilcox and Dopico (2011) extended the empirical evidence on cost efficiency by considering credit union mergers and efficiency; their study provides evidence using simple descriptive analysis to support the case that mergers tend to improve credit union cost efficiency. In situations where the acquirer is much larger than the target credit union, target members benefit in terms of lower loan rates and higher deposit rates, while acquirer members see little change as cost benefits are shared equally for when the merging credit unions are equal in size. 72 University of Ghana http://ugspace.ug.edu.gh Wheelock and Wilson (2013) attempted to assess how credit unions in the United States have controlled cost over time using an adapted version of the ‘‘order-a quantile’’ frontier estimate with the plausible effect of lowering loan rates to member-borrowers and high deposit rates to member-savers. Comparing larger to smaller size credit unions, the authors show that small size credit unions confronted a shift in technology that increased the minimum cost required to produce given amounts of output and that all but the largest credit unions also became less scale efficient over time. From Fukuyama, Guerra, and Weber (1999), we gain insight into how ownership affects efficiency in a cooperative setting. Providing empirical evidence that foreign-owned Japanese cooperatives mostly owned by Koreans are more efficient and experienced greater productivity growth during the period 1992 to 1996, they further argue that the history of institutional discrimination against Koreans in Japan suggests that ownership might affect overall efficiency using DEA. Servin, Lensink and Berg (2012) used stochastic frontier analysis to examine technical efficiency of different types of microfinance institutions including credit unions in Latin America and tested whether differences in intra- and inter- firm technical efficiency can be explained by differences in ownership. The evidence from this study is that non- governmental organizations and cooperatives have much lower inter-firm and intra-firm technical efficiencies than non-bank financial intermediaries and banks, indicating the importance of ownership type for technical efficiency. 73 University of Ghana http://ugspace.ug.edu.gh Glass, McKillop, Quinn, and Wilson (2014) demonstrate that the cooperative bank sector is characterized by increasing returns to scale as a result of mergers taking place. For example, Japanese cooperatives over the period 1998 to 2009 have secured both technical progress and a decrease in technical inefficiency. However, the empirical evidence from these cooperatives serve as a caution that regulatory pressure to decrease non-performing loans will have an unfavorable impact on both output and performance. Wijesiri, Viganò and Michele (2015) provide evidence from Sri Lanka on technical efficiency and its determinants for microfinance institutions including credit unions. In their second stage regression, age and capital-to-asset are significant determinants of financial efficiency whereas age, institution type and return-on-asset are the crucial determinants of social efficiency. 2.7 Conclusion and Research Gap The literature review has advanced reasons why financial institutions, including credit unions, may want to diversify revenue streams, the literature has presented varying conclusion on the effects of non-loan income on financial performance of banks and credit unions. The evidence show that the main effects are either positive or negative with a few works advancing arguments that non-loan income does not have a significant effect on financial performance. The inconclusive nature of this relationship gives much room for further studies to be conducted on credit unions, especially from the emerging market context as the current 74 University of Ghana http://ugspace.ug.edu.gh literature sways in favor of empirical research from the developed markets. Further from the literature search conducted for this study, no study has focused on non-loan income diversification, the level of diversification that exist in the credit union and the drivers of non-loan income diversification within the credit union, what come close is the study of Goddard, McKillop and Wilson in 2008 which focuses on diversification and financial performance of US credit unions. Lending, which is a major role of deposit taking financial institutions, is conducted on a more relational basis in the credit union setting compared to the transaction based option mostly employed in banking industry. There is a dearth of empirical literature on the lending business in the credit union literature; consequently, this study identifies and seeks to provide some empirical evidence regarding the interplay of credit union specific factors and banking sector development on credit union loan portfolio. What come close to the issue of lending in a developing market context is Adusei and Appiah, (2011) who concentrated on determinants of group lending in the credit union industry in Ghana, using logistic regression on cross-sectional data for 2008. From this, an in-depth analysis into lending in the credit union using panel data is critical and useful as the current empirical literature does not address the discretional and non-discretional factors that influence credit union lending. On efficiency, while there is not much disagreement in the literature regarding parametric and non-parametric approach to estimating efficiency. There seems to be some level of 75 University of Ghana http://ugspace.ug.edu.gh non-agreement with regard to the efficiency scores estimated which is believed to be influenced by the efficiency technique used. This thesis makes the attempt to estimate efficiency with the Tone’s efficiency estimation technique for a not-for-profit organization like the credit union a technique that has little empirical evidence in the credit union literature. There exists a dearth of evidence from the application of the Tone’s version of the DEA technique in evaluating cost efficiency especially in the credit union literature, a gap that this study seeks to fill. We also consider technical efficiency which has been relatively neglected in the credit union literature. The various thematic areas for this thesis namely, income diversification, lending, cost efficiency and technical efficiency is presented in an integrated model as captured in Figure 3.1 and estimated empirically. The integrated model explains that credit union factors are internal and external, these factors influence the inputs: labour, deposit and non-current asset employed to produce outputs that is loan and non-loan activities. These two activities leads to loan income and non-loan income. The non-loan income activities includes non-financial income and liquid-financial investment. In transforming the inputs into outputs, efficiency becomes paramount in the transformation process, if the credit union is to report good financial performance measured in risk adjusted return on asset and risk adjusted return on equity. 76 University of Ghana http://ugspace.ug.edu.gh Credit Union Environmental Factors Inte rnal Factors External Factors Inputs Production or Outputs Transformation Process Loan Non-Loan Efficiency Labour Deposit Non – Current Assets Liquid – Financial Investment Non-Financial Loan Income Income Financial Performance (Risk Adjusted Return on Asset, Risk Adjusted Return on Equity) Figure 2.1: Credit Union Integrated performance model. Source: Author’s Conceptualization 2017. 77 University of Ghana http://ugspace.ug.edu.gh From these reviews, we present an integrated risk adjusted performance empirical evidence as conceptualise in Figure 3.1, on credit union financial performance, which provides new insights into our understanding of the credit union. 78 University of Ghana http://ugspace.ug.edu.gh CHAPTER THREE METHODOLOGY 3.1 Introduction The research methodology used to generate results for the analysis in order to achieve the stated objectives in this thesis is described in this chapter. Discussed here are the research design, the population of the study, sampling technique and sampling size. The empirical models for each objective, variables and variable descriptions are also presented. The control variable for each object is described. 3.2 Research Design The major issues of interest in this study include income diversification, lending, efficiency and how these factors overall influence credit union performance. These objectives demand the use of descriptive, test and discuss relationship between variables and undertake some causal estimation. Being effectively able to do this would require employing quantitative research methods using quantitative data. According to Burns and Grove (1993), quantitative research is a formal, objective, systematic process to describe and test relationships and examine cause and effect interactions among variables. In terms of modeling credit union performance we follow Goddard, McKillop and Wilson (2008) and adapt the model of Worthington to study cost efficiency and technical efficiency in the credit union setting. 79 University of Ghana http://ugspace.ug.edu.gh This is a longitudinal panel study as the same set of credit unions are studied for the sampled time period: 2008 to 2014. The use of longitudinal study would make it possible to determine variables patterns that exist in the credit unions over time. Further longitudinal study helps to study the patterns of causal relationships over longer time spans. We make use of secondary data collected from the CUA Ghana on credit unions, top 3 bank asset concentration calculated from the data from the Ghana Association of Banker’s annual database, inflation and treasury bill data from the Bank of Ghana, gross domestic growth from the World development indicator data base, top 5 bank asset concentration, banking sector Zscore and banking sector overhead cost to asset from the Global Financial Development database. Table 3.1 gives a breakdown of the 61 credit unions used for this study. Table 3.1: Chapter Distribution of Data set Chapter Community Parish Workplace Total Chapter Total Weight Ashanti 0 0 3 3 76 4% Brong Ahafo 3 0 3 6 35 17% Central 0 0 5 5 36 14% Eastern 3 0 5 8 35 23% Greater Accra 1 6 9 16 118 14% Northern 1 0 1 2 20 10% Upper East 0 0 4 4 14 29% Upper West 0 0 2 2 12 17% Tema 1 2 1 4 43 9% Volta 0 0 4 4 20 20% Western 3 0 4 7 26 27% Total 12 8 41 61 435 14% From Table 3.1, workplace credit union dominates our data set followed by community owned and the parish owned credit union 80 University of Ghana http://ugspace.ug.edu.gh 3.3 Population of Study The study population comprised all 435 credit unions registered under the umbrella organization CUA Ghana as the end of July 2017. The detail by way of chapter distribution and types is presented in Table 1.2. We initially intended to study all credit union in Ghana; this was not possible because of non-availability of data on some credit unions. To circumvent this data challenge, we weighted the number of credit unions in each chapter into the total credit unions to have a representative sampled selected for this study. This we could not pursue because of the non-availability of data for some credit unions in some years. The final 61 credit unions used had consistent data from 2008 to 2014. Furthermore a close look at the distribution of credit unions used for this study shows that the chapters have a fair representation in our sampled credit unions for the study. Finally a sample size of 14 percent of the credit union population and the details in Table 3.1 fairly represents the credit union industry in Ghana. The issue of survival bias is negligible in our sample because, the excluded credit unions are as a result of non-availability of their data at CUA and not because they had ceased to operate because of bad performance. 3.4 Sampling Technique and Sample Size We started our data search on all credit unions in Ghana. However, this was not successful because of data unavailability on some credit unions, and we had to eleminate these credit unions out from our data collection. After detailed examination of the available data at 81 University of Ghana http://ugspace.ug.edu.gh CUA, we settled on annual reports data for 61 credit unions for a 7 years period resulting in a balanced panel data set of 427. The distribution of the 61 credit unions which is 14 percent of the industry, does not significantly bias the conclusions as the data sample covers all the 11 regional chapters making up the CUA Ghana. The panel dataset used permits us to control for variables like differences in business practices across the sampled credit unions which account for firm specific heterogeneity. 3.5 Data Sources Data on credit unions is sourced from the Ghana Co-operative Credit Union Association (CUA) Limited. The data set for this study is secondary. Inflation and Treasury bill data was sourced from the central bank; Bank of Ghana of Ghana time series data base. Data for the calculation of top 3 bank asset concentration is from the Ghana Association of Bankers (GAB) and data on Gross Domestic Product Growth (GDP) is from the World Bank Development Indicator (WDI) Series. Financial development indicators such as Bank Asset concentration for top 5 banks and Bank Overhead cost to Total Asset are sourced from Global Finance Development (GFD) data base. 3.6 Measuring Income Diversification Because diversification income attempts to measure the varied nature of income for a financial institution, we deviate from using ratio of non-interest income to operating income as a measure of income diversification used by Goddard, McKillop and Wilson (2008) and use the Herfindahl Hirschman Index HHI which is used to measure competition and 82 University of Ghana http://ugspace.ug.edu.gh concentration in the extant finance and economic literature. The use of the HHI gives a clue about the spread in the non-loan income sources for a credit union. Lower HHI implies a good spread in income sources hence a high level diversified income source whiles high HHI indicates concentration on a few sources of income, hence low diversification in income sources. We adapt the diversification measure of Mercieca et al (2007). In this regard, we construct three Herfindahl Hirschman Index HHI measures for each credit union to account for diversification between non-financial income, liquid financial investment activities, and combined non-loan income. We construct a measure of diversification income within the non - financial income category as, 2 2 2  ENT  COM   OTI  HHINFI           (3.1)  NFI   NFI   NFI  The motive for estimating equation 3.1 is to capture the extent of diversification income within non-financial income category for the sampled credit unions to know the extent spread within this income group. Where HHI  is a measure for non-financial income FII diversification, NFI is non-financial income. Where ENT is income from entrance fee, COM is revenue arising from commission income and OTI is other non-financial income. From the above, higher values means less diversification within the non-financial income activities, lower values means more diversification among the non-financial income generating activities. 83 University of Ghana http://ugspace.ug.edu.gh Similarly, we construct the combined non-loan income diversification index which measures the income generated from all non-loan income activities of the credit union given as 2 2 2 2 2  LFI   OFI   ENT   COM   OTI  HHICOMB                (3.2) TNLNI  TNLNI  TNLNI  TNLNI  TNLNI  In equation 3.2, HHI  diversification index for combined non-loan income for the credit COMB union is an attempt to account for credit union income diversification with non-loan income activities. LFI is liquid-financial income, OFI is other liquid-financial income and TNLNI is total non-loan income. An increasing HHI means the credit union is concentrating COMB on one type of non-loan income and less of diversifying income from the other non-loan income generating activity. The reason for estimating equation 3.2 is to capture the extent of diversification income within non- loan income category for the sampled credit unions to know the extent spread in this income group. Finally, the study endeavors to capture liquid-financial investment HHI . This measure LFI  accounts for diversification within credit union liquid-financial investment for each credit union. Liquid-financial investment is used as a proxy for income generated from financial investment as the breakdown of liquid-financial income could not be gleaned from the financial statements; a composite sum, however, was presented in the account of the credit union. A breakdown of income generated from liquid-financial investment would have been much preferred. Nonetheless the liquid-financial investment suffices as a good proxy for income from liquid-financial investment. The proxy is computed as: 84 University of Ghana http://ugspace.ug.edu.gh 2 2 2  BANK   CUACCF   T .BILL  HHI      LFI       LIQUIDINVEST     LIQUIDINVEST      LIQUIDINVEST  (3.3) 2  OFININVEST      LIQUIDINVEST   The motivation for estimating equation 3.3 is the capture the extent of diversification within the liquid-financial investment activity of the credit union non loan income generating activities. In equation 3.3 HHI  is a diversification measure from liquid-financial LFI investment and BANK is investment in banks that earns interest, CUACCF interest earning savings in CUA central savings facility, T.BILL is treasury bill investment and OFINVEST is other type of financial investment that earns interest in other financial institutions besides banks. LIQUID INVEST is total liquid investment by the credit union for a year. An increasing HHI means the credit union is concentrating on one type of financial LFI investment and less of diversifying income from other liquid- financial investment available on the financial market. 3.7 Financial Performance of the Credit Union The first step to achieve the study’s stated objective is to establish the factors that influence financial performance of credit unions. Financial performance of the credit union in this study is Risk Adjusted Return on Asset ( RAROA ) and Risk Adjusted Return on Equity ( RAROE ). In a functional form, credit union performance is given as, Y  f X ,O (3.4) 85 University of Ghana http://ugspace.ug.edu.gh In equation 3.4, Y is financial performance, X is a set of internal variables and O is set of external variables that influence the performance of credit unions. The empirical financial performance model for the credit union, we specify following the work of Goddard, McKillop and Wilson (2008) is 12 3 Yit 0  DVIN  I  E     (3.5) 1 it j it k t it it j1 k1 The model in equation 3.5, holds the view that the financial performance of credit unions is influenced by a set of internal factors, and external factors and some other factors that cannot be accounted for in our model as captured by the error term. In this model specification, Y is a measure of financial performance; risk adjusted performance namely i ,t ROA RAROA that is risk adjusted rate of return on assets  and also RAROE ; ROA ROE risk adjusted rate of return on equity  , 0 is a constant, u i Credit union fixed ROE effect, 1 is the estimated coefficient for diversification index. The variable, DVIN it = HHI  and HHI  is diversification measure from liquid-financial investment, LFI NFI HHI  is a measure for non-financial diversification income and is a NFI HHICOMB diversification measure for the combined non-loan income of the credit union respectively. The I is a vector of internal variables namely, SIZE is natural logarithm of total asset. LATA is liquid financial asset to total asset. COTI is cost to income which is a measure of resource usage and management quality. LOTA refers to loan to total asset. BDLN represents non-performing loans to total loans. NWTA is the net worth to total asset, NIM is 86 University of Ghana http://ugspace.ug.edu.gh net interest margin, Zscore is used to measure solvency risk. OWNS is a dummy variable equal to one for the type of credit union, that is EMPL is employer owned, PARI is parish based, COMTY or community owned otherwise zero. BKCN Top 3 bank asset concentration to measure banking industry concentration. The E is a vector of external variables namely, INFL is annual INFL to measure general increase in price levels. The annual change in gross domestic product is captured as GDP .  it is the stochastic error term. From the literature review on diversification, it is clear that firm diversify their activities and income sources for various reasons including, stabilizing income and reducing over variability income. Also from the literature, factors that affects firms diversification activities can be internally driven or some time externally driven. In view of this the second step is to estimate the drivers of credit union diversification. The specified functional equation for diversification is given as DVIN  f I , E (3.6) Where DVIN is diversification index, I is a set of internal variables and E is a set of external variables that influence credit union diversification income. 12 3 DVIN it   I  E     (3.7) j it k t t it j1 k1 Where DVIN it = HHILFI  HHI  , and HHI  , HHI  is diversification index from NFI COMB LFI liquid-financial investment, HHI  is a diversification index for non-financial income and NFI HHI  is a diversification index for combined non-loan income as specified in equation COMB (3.1), equation (3.2) and equation (3.3) respectively. Where I it represents the vector of the 87 University of Ghana http://ugspace.ug.edu.gh internal variables under the control of the management of the credit union, including SIZE, LATA, LOTA, ZSCORE, COTI, NIM, LLP, NWTA, AGE, OWN, Eit is the vector of the external variables to which the management of the credit union must adapt. These are made up of BKCN, INFL and GDP,  it measure the individual effect, and the stochastic error term is it . We assume that the errors are independently and identically distributed i.i.d  , and their distribution is unknown. The Z-score is used to measure credit union financial stability, higher profitability and capitalization implies credit union stability. The Z-score measure is given as ROA E /TA Z  score  (3.8) ROA ROA is return on asset, E /TA is equity to total asset ratio, a higher ratio indicates higher risk-adjusted profits.ROA is the sample period sigma Return on asset for the 7 year period. From the economic stance the Z-score measure the probability of a credit union becoming insolvent when the value of asset falls below the value of debt. Higher Z-score indicates lower probability of insolvency risk and lower Z-score points to higher probability of insolvency risk. The detail of the variables used is presented in Table 3.2. 88 University of Ghana http://ugspace.ug.edu.gh Table 3.2 Income Diversification and Financial Performance estimation variables. Variable Description Source Role RA DV ROA Risk Adjusted Return on Asset CUA RA DV ROE Risk Adjusted Return on Equity CUA HHI Hifindal Hirschman diversification measure from liquid-  DV,IV LFI  CUA financial Investment income HHI Hifindal Hirschman diversification measure for non- DV,IV NFI  financial income. CUA HHI Hifindal Hirschman diversification measure for combined   CUA DV,IV COMB non-loan income SIZE Total asset in natural log. CUA IV LATA Liquid Asset to total asset (percent) CUA IV LOTA Loan total asset (percent) CUA IV ZSCORE Insolvency risk; ROA is the standard deviation of ROA IV for credit union. CUA AGE Number of Years CU has been in existence CUA IV COTI Cost to income (percent) CUA IV NIM Net Interest Margin (percent) CUA IV LLP Loan loss to Loans (percent) CUA IV NWTA Net worth to total asset (percent) CUA IV BKCN Bank asset concentration for top 3 banks (percent) GAB IV INFL INFL rate (percent). BoG IV GDPG Gross Domestic Product growth (percent). WDI IV EMPL Employer based Credit Union (Dummy, 0,1) CUA IV Credit Union based on Community membership IV COMTY (Dummy, 0,1) CUA Credit Union based on religious believe IV PARISH (Dummy, 0,1) CUA IV is Independent Variable, DV is dependent Variable 3.7.1 Income Diversification Estimation Issues We employ three econometrics estimation models to investigate the problem under study. The Random Effect model is used in estimating model specified in equation (3.7) because the variation across credit unions is assumed to be random and uncorrelated with the predictor or independent variables included in the model. Furthermore, in estimating the risk adjusted return differences across credit unions is considered to have some influence on the risk adjusted return dependent variable in our model and in estimating the model specified 89 University of Ghana http://ugspace.ug.edu.gh in equation (3.5), which includes time invariant variables such as ownership type of the credit union. Again the Hausman and Breusch-Pagan Lagrange Multiplier (LM) Test were used to confirm the choice of estimation. In estimating the model specified in equation (7), the Hausman and Taylor (1981) instrumental variable estimation is used to accommodate the fact that in the random-effects models some of the covariates are correlated with the unobserved individual-level random effect. It is further assumed that some of the explanatory variables are correlated with the individual-level random effects, u i but that none of the explanatory variables are correlated with the idiosyncratic error, it . We judge the endogeneity of size recommended by Berger and Udell (2002) and include size as endogenous. Cost to income is also considered as endogenous in the Hausman and Taylor (1981) estimation for non-financial income. In respect to liquid-financial investment, NWTA and NIM serve as the instrumental variables. For combined non-loan income NIM and GDP are used are endogenous variables. Finally the Tobit estimation method is considered to estimate the model specified in equation (3.5) because the diversification index value ranged from 0 to 1. The Tobit model, also called a censored regression model, is used to estimate the relationship between the dependent variables revenue diversification and the explanatory variables. In Tobit estimation method the dependent variable is either censored from the left, also called from below, or censored from the right, also called from above. Censoring from above takes place when cases with a value at or above some threshold all take on the value of that threshold, so that the true value might be equal to the threshold, but it might also be higher. In the case 90 University of Ghana http://ugspace.ug.edu.gh of censoring from below, values that fall at or below some threshold are censored. Due to the spread of diversification index value from 0 to 1, we censored from the left, a lower limit of 0.45, 0.46 and 0.677 for non-financial income, liquid-financial investment and combined non-loan income in the dependent variable respectively. See appendix 2a, 2b and 3c for the histogram on the index and the choice of lower level limit choose in the Tobit estimation. From this equation 3.7, becomes, a if x '    a,    y  b if x '    b, (3.9)    ' x    otherwise, In this censored Tobit formulation y is the observed value of the dependent variable, x is a vector of observed explanatory variable,  is a vector of unknown regression coefficients to be estimated, is an unobserved error term and a and b are the censoring limits. 3.8 Credit Union Lending In modeling the loan business of credit unions, we express loan granted as a function of internal and external determinants. The internal determinants are discretional factors that management of the credit union can control; in this study it is composed of the pecuniary items derived from credit union annual reports (the income statement and statement of financial position) which are termed credit union specific determinants of loans. The external determinants are non-discretional variables that the manager of the credit union must adapt to since, as documented in the literature, these factors tend to relate to the 91 University of Ghana http://ugspace.ug.edu.gh lending business of financial institutions, and reflect the industry and economic environment that affects the operation and performance of financial institutions. We also reiterate that the choice of credit union specific and financial development variables employed in our empirical model is heavily influenced by the theoretical works of Smith (1984 and 1988) and Rubin, Overstreet, Beling & Rajaratnam (2013). Credit union lending in this study is loan to total asset ratio. Loans are created from excess reserves of the credit union. Further the excess reserves are decomposed into loan and non- loan portfolios, namely financial investments and non-interest earning income generating activities. It is obvious from the composition of excess funds that an indirect relationship exists between loan portfolio and non-loan portfolio as an increase in loan extended during a period would mean less funds available for non-loan activities. We assume that there is a no change in funds mobilized from depositors and there exists a constant demand for loans and constant demand for funds for non-loan activities. The study used the combined diversification measure informed by the works of Mercieca, Schaeck & Wolfe (2007) and for the purpose specified in equation 3.3, to account for income diversification from non- loan income activities. We reiterate that an increasing HHI means the credit union is concentrating on one type of non-loan income and less COMB on diversifying income from the other non-loan income generating activity. 92 University of Ghana http://ugspace.ug.edu.gh 3.8.1 Credit Union Lending Econometric Estimation The study estimates two econometric models, the fixed effects model or the random effect model as specified in equation 3.9, for fixed effect the reason is that we are interested in variables that vary over time. The fixed effect estimation is used because it is assumed that some credit union specific variables may influence the predictor variables and this need to be controlled for. The author acknowledges that in estimating fixed effects model, time invariant cannot be explored because of credit unions’ loan to asset ratio. In the Random Effect model estimation, the variation across the credit unions sampled for this study is assumed to be random and uncorrelated with the explanatory variables in the model. Additionally, the differences across credit unions is held to have some influence on the loan to asset ratio hence the use of random effects. Finally Random Effect permits the generalization of inferences beyond the sampled credit unions used in this study. 93 University of Ghana http://ugspace.ug.edu.gh Table 3.3: Variable in the Lending Model. Variables Description Source LNLOTA Natural log Loan to total asset (percent) CUA SIZE Natural log of Total asset CUA SIZE SQUARE Square of natural log of Total asset ROE Net Income scaled by total equity (percent) CU A LLP Loan loss to Loans (percent) CUA NWTA Total Asset less Liabilities / Total Asset (percent) CUA NIETA Net Interest Expense to total asset (percent) CUA Hifindal Hirschman diversification measure for combined non-loan HHI COMB income ZSCORE Insolvency risk;ROA is the standard deviation of ROA for credit union. CU A AGE Number of years credit union has been operational CUA LENDRATE Interest Rate Charge on member loans (percent). CUA BACN Bank asset concentration for top 5 banks (percent) GFD BOCTA Bank overhead scaled by Total Asset GFD TBILL 1 year Treasury Bill rate (percent) BoG GPDG Gross Domestic Product growth (percent). WDI CUA is Credit Union Association Ghana, GFDD, Global Finance Development Data Base, BoG is Bank of Ghana. In the lending business of the credit union, internal and external factors tend to influence the loan portfolio created by the credit union. The panel empirical model we specify in the general form is, 14 2 2 LNLOTAit 1  jCUSVit  k FDI t k MACRt  (3.10) it j1 k1 p1 Financial development indicators ( FDI ) is made up bank asset concentration for top 5 banks, and bank overhead cost to total asset. The error term from equation 3.9 is given as,  it  vi  t  eit (3), where vi the individual is credit union effect t is the between entity error term associated with time and eit is the within entity error. The estimation of the model; Random Effect or Fixed Effect is decided using the Hausman Test. 94 University of Ghana http://ugspace.ug.edu.gh In the model specified in equation 3.9 LNLOTA is credit union lending, which is the loan to it asset ratio,  is a constant, CUSVit is a vector of credit union firm specific variables including, size, return on equity, loan loss to total loan, net worth to total asset, net interest expense to total asset, diversification income, Zscore which measures the solvency level of the credit union, age of the credit union and lending rate a measure of rate charge on loans extended by the credit union. The FDI t is a vector of financial development indicator variables namely top 3 bank asset concentration a measure of banking sector competition and bank overhead cost to total asset and indicator for banking sector efficiency. MACRt is a set of macroeconomic variables including inflation adjusted 1 year Treasury Bill rate, a measure of riskless investment available on the Ghanaian market that is a proxy for monetary policy decision by the central bank. GDPGannual gross domestic product measured in growth GDPG a measure size of economic growth,  it is the stochastic error term. 3.9 Credit Union efficiency There are convincing reasons in favor of both parametric and nonparametric approaches to estimating cost efficiency and technical efficiency. For the purpose of this study, the non parametric DEA technique is chosen because it does not necessitate the specification of arbitrary functional form so a possible misspecification of a parametric function is avoided. In addition to the strengths of DEA presented in the literature review, the work of 95 University of Ghana http://ugspace.ug.edu.gh Worthington (2000) based on the DEA technique lends support to this technique employed in our study on the credit union. We also add that DEA technique was best suited for a not-for-profit organization like the credit union and also the technique fits the original organization for which the DEA was formulated for evaluating performance in non for profit organization although we do admit that other authors have used other techniques such as the Stochastic frontier analysis in evaluating efficiency. For example, in his 2001 assessment of Australian finance cooperatives, including credit unions, Esho uses stochastic frontier analysis while McKillop et al (2002) employ the radial and non-radial approach on UK credit unions and Wheelock and Wilson (2013) use a cost analog of the Malmquist productivity index in assessing cost- productivity and efficiency in US credit union. The works of Koopmans (1951), Debreu (1951), and Shephard (1953) provides an important basis for models that have assessed frontier efficiency and productivity of decision making units, be they profit or non-profit making DMU. In particular, Debreu and Shephard employed distance functions to model multiple outputs on one hand and to measure radial distance of a producer from the frontier on the other hand. The pioneering DEA model in the efficiency literature is by Charnes, Cooper and Rhodes, known as the CCR model, after its authors. Readers interested in the mathematics of the DEA are referred to the works of Charnes, Cooper and Rhodes (1978), Banker et al (1984), 96 University of Ghana http://ugspace.ug.edu.gh Coelli (1996), Worthington (2000), Camanho and Dyson (2005), Sufian (2009), AlKhathlan and Abdul Malik (2010) Fujii, Managi, Matousek, and Rughoo (2017), among others. Very few studies have however adopted Tone (2002) cost-efficiency model in banking or openly discussed it (Tone and Sahoo 2005, Tone and Tsutsui 2007, Dong, Hamilton et al. 2014). The original data envelopment analysis (DEA) by Charnes, Cooper et al. (1978) which assumes constant returns to scale have been extended to the variable returns (VRS) (Banker, Charnes et al. 1984), profit maximization Fried, Schmidt and Lovell (1993) and cost minimization (Färe, Grosskopf et al. 1985).Still, the basic CCR and BCC models measure technical efficiency (Dong et al. 2014). Yet, this stduy focus on measuring the cost efficiency (also referred as overall efficiency in the literature).Cost efficiency assesses the ability to produce current outputs using minimal input mix. It is the product of allocative and technical efficiency (Fethi and Pasiouras 2010). Allocative efficiency examines the ability of a credit union to use optimal mix of inputs given their prices whilst technical efficiency refers to the ability to use the right resources appropriately. Hence, cost efficiency seeks to avoid resource wastage resulting from technical or allocative inefficiency. The cost efficiency of observed credit unions is given by the distance of its detected cost point from a constructed cost frontier (Dong et al., 2014). Thus, the cost efficiency (CE) of an evaluated credit union is modelled as 97 University of Ghana http://ugspace.ug.edu.gh wij x i j CE  (3.11) wij xij The ideas of cost efficiency were pioneered by Farrell (1957b) and Debreu (1951), and later developed into implementable approach by Färe et al. (1985a) using a linear programming technique (Camanho & Dyson, 2008; Dong et al., 2014).The traditional cost efficiency DEA model by Färe, Grosskopf, and Lovell (1985a) assumed that input prices are the same across all DMUs, though actual markets may function under imperfect competition and unit input prices may not be the same across all DMUs. Tone (2002) identified these drawbacks in the conventional cost efficiency DEA model and henceforth proposed a new scheme to measure cost efficiency. The new cost efficiency was further extended to decomposition of cost efficiency by Tone and Tsutsui (2007). Under the new cost efficiency model DMUs with dissimilar input prices will provide different measures of cost efficiency (Dong et al., 2014). The traditional (Färe et al., 1985a; Farrell, 1957a) and new cost efficiency DEA (Tone, 2002) model are expressed in equations (2) and (3) respectively.  min , xi wiO xio , (3.12) 98 University of Ghana http://ugspace.ug.edu.gh Where denotes the cost minimizing vector of inputs quantities for the credit union under observation, given the vector of output weights and input prices . Hence, a hypothetical credit union with the same input price vector as the observed credit union , will have a cost efficiency score of .   min , xCT  e x  , (3.13) iO Where e ϵ a row vector with all elements equal to 1 and T. The Tone (2002) cost model differs from the traditional model because under the latter, observing the credit union unit cost of credit union j fixed at , we search for the optimal input mix for producing . However, under the former the optimal input that 99 University of Ghana http://ugspace.ug.edu.gh produces the output can be found independently of the credit union’s prevailing unit price . Thus, based on the optimal solution Tone (2002) cost efficiency (CET) is defined as _ e x  CET  iO (3.14) _ e x This method is utilised because VRS relaxes the constant returns to scale assumption and allows for the possibility that the credit unions’ production technology may exhibit increasing, constant, or decreasing returns to scale. Again the input-orientated DEA model helps to analyze how much a credit union input quantities can be proportionally reduced without changing the output quantities produced. As credit union management have no control over input cost, the input-orientation we believe is best suited for our objective. For technical efficiencyTE , in the input-oriented value – based framework is set up as TE  Min   , subject to : j xo  x io , j 1,...,m jJ  j yrj  yro , r 1,..., s, jJ (3.15)  j 1,  j  0, j J. jJ From equation 3.15, , is a set of variables , that input and output variables, x , is a set of inputs , y , is a set of outputs. The subscript j is the set of inputs for credit unions 1 to mth credit union. The subscript r is the set of output from credit union 1 to the sth credit union. 100 University of Ghana http://ugspace.ug.edu.gh 3.9.1 Inputs and Outputs Specification In defining the inputs and outputs in financial institutions, the approaches used in literature include the intermediation approach, the production approach, the asset approach, the user cost approach, and the value-added approach. Of these, two prominent approaches emerge: the intermediation approach and production approach. Berger and Humphrey (1997) use the Production Approach to present the financial institution as a producer of deposit accounts and loans, the output is the number of accounts or their associated type of transactions including deposit and loan accounts, and inputs are calculated as the number of employees (labour) and capital expenditures on noncurrent asset and other material. Using the Intermediation Approach, Sealey and Lindley (1977) view financial institutions as intermediaries, converting and using factors of production paid for by owners to transfer financial saving from surplus units to deficit units in the form of loans. The inputs may include labour and capital costs, and the interest payable on deposits, with the outputs denominated in loans and financial investments among others. The intermediation approach is preferred over the production approach for the current study because, although the credit union conducts business for owner-members, the manager still intermediates between member borrowers (deficit units) and member savers (surplus units). For the purpose of this study we hold the assumption that the credit unions makes use of three inputs to produce three outputs. The three inputs consumed by the credit union namely 101 University of Ghana http://ugspace.ug.edu.gh Labour, Non-current Asset and Deposits; the three outputs are Loan, Liquid Financial Investment and Non-Loan Income. Because these inputs are not free but comes at a cost, three input prices are estimated; these are price of labour, price of noncurrent asset and price of funds. The cost of labour is proxied by personnel costs. The study would have benefitted from data on fulltime and part time employees to be able to estimate an exact cost of labour; these were unavailable. Nonetheless, we believe the current approach can serve as a good proxy for cost of labour. Table 3.4 gives further details on the inputs, outputs and input price. 102 University of Ghana http://ugspace.ug.edu.gh Table 3.4 : Efficiency and Regression Variables Variables Description Source Inputs Labour Total expenditures on employees (personal expenses). CUAFS Noncurrent Asset Non Current Asset. CUAFS Deposits Total deposits mobilized. CUAFS Outputs Loan Total Loan. CUAFS Liquid Financial Sum of interest income from liquid financial CUAFS Investment investment. Non-Interest Sum of non interest income. CUAFS Income Input Price Price of labor Total personal expenses scaled by the total funds. Author’s own calculation Price of Non Depreciation expenses scaled by Non Current Asset. Author’s Current Asset own calculation Price of funds Interest expenses on deposits and non-deposits funds Author’s plus other operating expenses divided own by the total funds. calculation Dependent  Cost efficiency score generated from the input oriented Author’s CE VRS Tone’s SBM DEA approach. own calculation  Technical efficiency score generated from the input Author’s TE oriented VRS approach. own calculation RA Risk Adjusted Return on Asset Author’s ROA own calculation RA Risk Adjusted Return on Equity Author’s ROE own calculation Explanatory SIZE Total asset in natural log. CUAFS 103 University of Ghana http://ugspace.ug.edu.gh ZSCOR Insolvency risk; σ is the standard deviation of ROA for Author’s credit union. own calculation NWTA Net worth to total asset (%). Author’s own calculation BDLN Loan loss to Loans (%) . CUAFS HHI  Hifindal Hirschman diversification measure for non- Author’s NFI financial income. own calculation HHI  Hifindal Hirschman diversification measure from Author’s LFI liquid-financial Investment income own calculation HHICOMB Hifindal Hirschman diversification measure for Author’s combined non-loan income. own calculation NIETA Net Interest Expense to total asset (%). CUAFS LOTA Loan total asset (percent). CUAFS AGE Number of years a credit union has been in existence. Author’s own calculation LITA Liquid Asset to total asset (%). CUAFS 104 University of Ghana http://ugspace.ug.edu.gh BKCN Asset of five largest banks as a share of total GFDD commercial banking asset. BKZS It captures the probability of default of a country's GFDD commercial banking system. It is estimated as (ROA+(equity/asset))/σ(ROA); σ (ROA) is the standard deviation of ROA. BOCTA Operating expenses of a bank as a share of the value of GFDD all asset held. RLTB Inflation Adjusted 1 year Treasury Bill rate %. BoG GDPG Gross domestic product growth %. WDI CUAFS is Credit unions’ annual financial statements, GFDD, Global Finance Development Data Base, BoG is Bank of Ghana. WDI, World Development Indicator Series. 3.9.2 Efficiency Data Source Firm specific data from the annual financial statement report of 61 credit unions operating under the association of CUA of Ghana were sampled. The period for this study is from the year 2008 to the year 2014. The panel data set used permits control for variables like differences in business practices across the sampled credit unions which account for firm specific heterogeneity. Banking industry indicators is sourced from Global Finance Development database, real Treasury bill rate is calculated from data provided by the Bank of Ghana and GDP growth rate is from the World Development Indicators. 3.9.3 Efficiency Econometric Estimation In examining the factors that could influence credit union efficiency, the two stage semi- parametric process is adopted. In the first stage cost and technical efficiency scores for all 105 University of Ghana http://ugspace.ug.edu.gh credit unions in the sample are estimated as specified in equation (3.14) and (3.15) using a set of inputs, input prices and outputs. In explaining cost efficiency and technical efficiency in the credit union, credit union factors, banking sector development and the macroeconomic development, from the empirical literature impacts on credit union efficiency. In the second stage, the efficiency estimates obtained are regressed on the selected internal and external variables as specified in equation (3.16) using mixed effects and a Two Limit truncated Tobit regression method. A two limit Tobit regression is also used because the cost and technical efficiency estimates is censored Lower and Upper values and no estimates fall outside this range. The use of Tobit regression is also justified by Hoff (2007) who makes the case that this regression technique is sufficient in representing second stage DEA models. In using these regression methods the embedded maximum likelihood treats the correlation challenge asymptotically albeit at a measured rate. The lower limit for cost efficiency is 0.1078 with upper limit of 1.000, for technical efficiency, the lower limit is 0.1102 and the upper limit is 1.000. We again use the formulation specified in equation 3.8 in our Two Limit truncated Tobit regression on the efficiency scores. 106 University of Ghana http://ugspace.ug.edu.gh  'a if xit    it  a,    ' it  b if xit    it  b, (3.16)    'xit    it otherwise, In this censored Tobit formulation it is the observed value of the dependent variable which is the estimated cost efficiency and technical efficiency score from the first state DEA, xit is a vector of observed explanatory variables presented in Table 3.4,  is a vector of unknown regression coefficients to be estimated, it is an unobserved error term and a and b are 0.1078 and 1 and 0.1102 and 1 for the cost efficiency and technical efficiency censoring limits respectively. The mixed effect models uses both fixed and random effects model. These correspond to a hierarchy of levels with the recurring, correlated measurement occurring among all of the lower level units for each particular upper level unit. The use of the mixed effects allows for inference on populations but not on individual credit union. The structure of the data for the study, makes it possible to group the data at the lower level that is the credit union type level and the upper level which is to group the credit unions into the 11 chapters. Hence we estimate the following mixed effects panel regression model, i, j ,k    eijk (3.17) 107 University of Ghana http://ugspace.ug.edu.gh From equation 3.16,  is the observed cost efficiency and technical efficiency estimates from the first stage DEA estimation, i is grand credit union level, j is the type of credit union, k is the observation. The  is the fixed effect part of the model, composed of explanatory variables made up of a set of credit union factors, banking industry variables and macroeconomic variables presented in Table 3.4. The  is model as 9 3 2  it 1 1CU Factors it 2 Banking Industry  t 3Macrot   it j1 k1 n1 The eijk is the random effect part of the model which is the residuals. In the model in equation 3.16,  , is the estimated cost efficiency and technical efficiency derived from the i ,t input oriented Tone’s cost efficiency, CU Factors is a vector of credit union specific characteristics, Banking Industry is a set of banking industry condition and Macro is a set of macroeconomics variables,  is the error term, and the subscripts i and t represent i ,t individual credit union and time period, respectively. The credit union firm specific factors are size of the credit union captured by natural log of total asset, LOTA is loan to asset ratio and LITA is used to capture liquidity of the credit union (see McKillop et al 2002). LLP is used to capture loan quality similar to Worthington (2000). However, instead of using noninterest income to total revenue as Worthington does, the c income diversification is captured using HHI as it is believed this would give a COMB good measure of income from non-loan activities. Furthermore combined non loan diversification measure is included because it is believed that this activity can expose the credit union to high level of risk hence high cost inefficiency if income diversification is not pursued cautiously. Age is used to capture the years of existence of a credit union (see Esho 108 University of Ghana http://ugspace.ug.edu.gh 2001). NWTA is net worth to total asset a measure of equity holder’s contribution. Zscore is used to capture solvency risk; NIETA is used to measure management control expenses. To measure the relationship between banking industry and macroeconomic conditions and credit union efficiency, the following variables in the regression model are employed: BKCN5 (the concentration ratio of the five largest banks in terms of asset) is to measure banking sector concentration; BKZS is Bank Zscore (Banking industry stability); BOCTA is bank overhead cost to total asset, a measure of banking sector efficiency. Also introduced are the GDP (gross domestic products) to control for cyclical output effects, and the RLTB (nominal one year Treasury bill rate adjusted for inflation), to control for monetary policy stance. 3.10 Integrated Performance of the Credit Union Finally an integrated performance model as specified in equation 3.18 is estimated in an attempt to capture income diversification as measured in this study, as well as loan to asset estimated cost, technical efficiency and other pertinent variables identified from the literature review on risk adjusted credit union performance. This model incorporates all the three main themes of this thesis, namely, income diversification, lending, cost efficiency and technical efficiency and their effects on risk adjusted financial performance of the credit union. In this regard a mixed effect model is specified as, 10 3 2 yit  1CU Factors it 2 Banking Industryt 3Macrot   it j1 k1 n1 (3.18) 109 University of Ghana http://ugspace.ug.edu.gh From equation 3.18, the error term,  it is decomposed as  it  U itCU Factorsit U t Banking Industryt U t Macrot In the integrated model the firm specific factors includes the measured non-loan diversification index, loan to asset, and the cost and technical efficiency scores. One distinction made in estimating equation 3.18 is that cost to income is replaced with cost efficiency in each risk adjusted performance model with technical efficiency as there are no input prices under estimating technical efficiency 110 University of Ghana http://ugspace.ug.edu.gh CHAPTER FOUR RESULTS AND DISCUSSION 4.1 Introduction This chapter presents the results of the analysis based on the methods discussed in Chapter Three. These results are discussed with the literature review in mind and are organised in sections based on the main objectives of this thesis. The chapter begins with combined presentation of the descriptive statistics in section 4.2, capturing all the variables used in this thesis. Section 4.3 presents the discussion on income diversification while section 4.4 focuses on lending. The case for cost and technical efficiency is made in section 4.5 and lastly we present the integrated performance discussion in section 4.6. The chapter ends with as summary of the hypotheses based on the results generated from the estimations in section 4.7. 4.2 Descriptive Statistics The descriptive statistics in Table 4.1 show that the risk adjusted returns for credit union sample from Ghana for the study hovered around an average of 1.805 for RAROA and RAROE . The sample mean are 0.178, 0.479 and 0.608 for HHI  , HHI and NFI LFI  HHICOMB respectively, with HHI  recording the highest variability of 0.234. NFI 111 University of Ghana http://ugspace.ug.edu.gh Table 4.1 Combined Variables Descriptive Statistics Variable Mean Std.Dev Min Max Dependent RAROA 1.8021 2.1542 (5.4760) 13.3765 RAROE 1.8106 2.1322 (5.5893) 12.8771 HHI NFI  0.1781 0.2346 0.0000 1.0000 HHI  LFI 0.4790 0.1660 0.2324 1.0000 HHI COMB 0.6048 0.2011 0.2254 0.9954 CE 0.3888 0.2149 0.1078 1.0000 TE 0.5439 0.2514 0.1102 1.0000 Independent SIZE 5.8300 0.5696 4.4019 7.5441 LNLOTA 1.7199 0.1393 0.9185 1.9585 LATA 30.6862 16.4454 0.9239 100.0000 LOTA 54.8063 14.8934 0.0000 96.8126 Z-SCORE 0.5138 0.3955 0.0220 2.6378 AGE 16.9485 10.6627 1.0000 45.0000 COTI 81.8940 33.5904 9.7015 374.6198 NIM 14.8184 10.1213 (20.1642) 162.1970 LLP 3.0343 5.4991 0.0000 59.0278 NWTA 94.5057 5.3340 54.9623 100.0000 NIETA 8.3552 4.5164 0.3810 39.7088 ROE 3.6543 7.4105 (27.8708) 122.4626 LDRTE 27.5766 2.2902 25.0313 31.9167 BKCN3 33.6600 3.3938 28.5460 38.6385 BKZ-SCOR 6.3229 0.8641 4.9300 7.5300 BKCN5 63.4629 10.5217 55.7100 87.3200 BOCTA 6.9400 1.0747 4.7800 8.0300 INFL 12.9429 3.9255 8.5800 18.1000 TBILL 18.8643 4.5193 11.3000 22.9000 RLTB 5.1071 3.8059 1.9000 14.0600 GDPG 8.0760 3.0751 3.9859 14.0460 N=427 , n=61 N: number of observations; n: number of credit unions The overall mean for the size of credit union is 5.8300, and on average 30 percent of the total asset were invested into liquid-financial asset. In terms of resource usage measured by the COTI, the cost to income was 81.89 percent of total income generated for the sample 112 University of Ghana http://ugspace.ug.edu.gh credit union. Top 3 bank asset concentration was 33.66 percent on the average, inflation had a mean value of 12.94 and average growth in GDP was 8.07 percent. The net interest expense to total asset has a mean value of 8.4310%. The range of values for diversification income measured by HHI  , ranges between 0.2254 to 0.9954 indicating COMB that during the period under study some credit unions concentrated on some non-loan income activities as inferred from the high diversification index value. Other credit unions also experienced a good spread of funds across non-loan activities, a conclusion drawn from the extremely low diversification of 0.2254. The average solvency captured by the Zscore was 0.5138 implying that most credit unions were fairly solvent during this period. The average interest rate charged on loans by the credit unions was 27.5766 percent with 31.9167 percent being the highest rate charged; this was gleaned from the loan interest income. Top 5 bank asset concentration over the period was 63.4629 implying a non-competitive banking sector in Ghana. Bank over cost to total asset ranged between 4.7800 percent to 8.0300 percent. The average rate was 18.8643 percent and 8.0760 percent for Treasury bill and GDP growth respectively. Loan to total asset was measure in natural log to smoothen out outliners, during the period under study, loan to total asset was 1.7199 on the average, with a total variation of 0.1393 and a minimum value of 0.91851 and highest value of 1.9585 113 University of Ghana http://ugspace.ug.edu.gh The descriptive statistics in Table 4.1 reveal that during the period under study some credit unions in Ghana were cost efficient scoring 1, with poor performers scoring cost efficiency of 10.78 percent with a mean value of 38.88 percent. For technical efficiency best performing credit unions score 1 and worst performers recorded 11.02 percent with an average efficiency score of 54.39 percent. On the average, most credit unions in Ghana were more technically efficient than cost efficient. The stability of the banking sector capture by Bank zscore was 6.3229 on the average during the period 2008 to 2014 Bank over cost to total asset was 55.5071 on the average with real Treasury bill rate recording 5.1071 percent on the average. The highest GDP over the period was 14.0460 percent which was recorded in the year 2011. Table 4.2 is the correlation matrix for the variables used in the income diversification objective of this study. 114 University of Ghana http://ugspace.ug.edu.gh Table 4.2: Correlation Matrix of Income Diversification Variables RAROA RA HHI HHI HHI ROE (LFI ) (NFI ) (COMB) SIZE LITA LOTA ZSCOR AGE COTI NIM LLP RAROA 1.00 RA ROE 1.00 1.00 HHI (LFI ) 0.01 0.00 1.0 0 HHI (NFI ) (0.12) (0.12) (0.18) 1.00 HHI (COMB) 0.10 0.11 0.38 (0.41) 1.00 SIZE 0.35 0.35 0.01 (0.13) (0.04) 1.0 0 LITA 0.08 0.07 0.22 (0.41) 0.20 (0.12) 1.0 0 LOTA (0.06) (0.06) (0.10) 0.25 (0.12) 0.09 (0.80) 1.0 0 ZSCOR 0.73 0.72 (0.01) (0.00) 0.07 0.21 (0.07) 0.01 1.0 0 AGE 0.26 0.26 0.16 (0.09) 0.25 0.45 (0.07) 0.09 0.17 1.0 0 COTI (0.58) (0.59) 0.01 0.04 (0.09) (0.24) (0.08) 0.04 (0.13) (0.20) 1.0 0 NIM 0.18 0.18 (0.18) (0.04) (0.21) 0.08 0.12 (0.22) (0.03) (0.05) (0.21) 1.0 0 LLP (0.28) (0.29) 0.08 (0.07) (0.08) (0.03) 0.14 (0.13) (0.14) (0.14) 0.44 0.10 1.0 0 NWTA 0.08 0.05 0.05 (0.02) (0.03) 0.08 0.07 0.08 0.11 0.03 (0.00) (0.01) 0.06 BKCN (0.07) (0.06) (0.07) 0.16 (0.05) (0.37) (0.12) 0.15 (0.01) (0.17) 0.06 (0.15) (0.05) INFL 0.02 0.03 0.00 (0.07) 0.08 (0.08) (0.04) 0.05 (0.01) (0.03) (0.05) (0.02) (0.08) GDPG (0.06) (0.07) 0.03 0.16 (0.14) (0.09) 0.01 0.01 0.00 (0.05) 0.10 (0.06) 0.07 LLP NWTA BKCN INFL GDPG L LP 1.00 NWTA 0.06 1.0 0 BKCN (0.05) (0.11) 1.0 0 INFL (0.08) (0.04) 0.36 1.0 0 GDPG 0.07 0.01 0.14 (0.61) 1.0 0 115 University of Ghana http://ugspace.ug.edu.gh 4.3 Income Diversification 4.3.1 Non-Loan Income Diversification From Table 4.1, we can deduce that credit unions sampled for this study, fairly diversified their income within the HHI  non-financial income group of activities, the mean value of NFI 0.1781 supports this assertion with a high standard deviation value of 0.2346, and also recorded some concentration on activities in the non-financial income set as seen in the maximum value of 1.0000 over the period. Within the liquid-financial investment HHI category, the mean value is 0.4790 depicting a fairly diversified investment source LFI among within this group although, there is evidence of over concentration on one source of liquid-financial investment income during the period for the study from the maximum value of 1.0000. The extent of diversification within the combined non-loan income group was relatively concentrated with an average value of 0.6048, implying that for the study period, credit unions relied heavily on some sources of non-loan income with little spread in this set of activities for non-loan income. The variability in these investments was relatively high at 0.2011 compared to 0.1660 for diversification within the liquid-financial investment category. Objective 1: Assess factors that impact overall credit union financial performance. The empirical model to assess the financial performance of credit unions was presented in equation 3.5. From the model we capture financial performance as risk adjusted return on 116 University of Ghana http://ugspace.ug.edu.gh equity and risk adjusted return on asset. The aim of this objective is to find out how diversification income from non-financial income, diversification activities within the liquid-financial investment and combined diversification income influence financial performance of credit unions. 4.3.2 Income diversification and Liquid-Financial Investment diversification: any role in Credit Union financial performance? Model Diagnostic tests We present empirical results for the financial performance model specified in equation 3.4. The model diagnostic test shows a significant Wald chi2 score in all the estimations in Table 4.2, the inference here is that all the parameters used in the model are jointly significant. The random effects model employed in model 1, 2 and 3 of Table 4.2 is based on the Hausman test with a significant probability value of more than 0.05. In model 4, 5 and 6, the Brusech -Pagan Lagrange multiplier (LM) show that the random effect is preferred as there is evidence of significant differences across the sampled credit unions. 117 University of Ghana http://ugspace.ug.edu.gh Table: 4.3 Financial Performance, Diversification Income and Investment Random Effects (1) (2) (3) (4) (5) (6) VARIABLES RAROE RAROA RAROE RAROA RAROE RAROA SIZE 0.2670** 0.2553** 0.2746** 0.2621** 0.2785** 0.2646** (2.1037) (1.9843) (2.1757) (2.0535) (2.2030) (2.0682) C OTI -0.0228*** -0.0225*** -0.0226*** -0.0223*** -0.0226*** -0.0223*** (-13.4790) (-13.3111) (-13.4133) (-13.2894) (-13.4088) (-13.2816) LOTA 0.0117** 0.0120** 0.0121** 0.0123** 0.0122** 0.0125** (2.1233) (2.1903) (2.2298) (2.2748) (2.2587) (2.3072) N IM 0.0144*** 0.0143*** 0.0149*** 0.0149*** 0.0152*** 0.0150*** (3.2091) (3.2078) (3.2937) (3.3182) (3.3531) (3.3389) L ATA 0.0180*** 0.0186*** 0.0186*** 0.0189*** 0.0187*** 0.0191*** (3.2781) (3.4019) (3.5366) (3.5892) (3.5855) (3.6681) L LP -0.0155* -0.0160* -0.0162* -0.0167* -0.0159* -0.0163* (-1.6666) (-1.7291) (-1.7399) (-1.8082) (-1.7101) (-1.7672) NWTA -0.0248*** -0.0142 -0.0260*** -0.0154* -0.0250*** -0.0143 (-2.6550) (-1.5306) (-2.7961) (-1.6596) (-2.6853) (-1.5395) ZSCORE 3.5576*** 3.6231*** 3.5560*** 3.6204*** 3.5458*** 3.6119*** (17.0619) (16.7537) (16.7359) (16.5099) (16.6242) (16.3812) BKCN -0.0009 -0.0035 -0.0012 -0.0032 -0.0014 -0.0038 (-0.0551) (-0.2187) (-0.0715) (-0.1943) (-0.0854) (-0.2351) I NFL 0.0028 0.0052 0.0023 0.0043 0.0031 0.0054 (0.1908) (0.3612) (0.1596) (0.2999) (0.2185) (0.3815) GDP -0.0120 -0.0093 -0.0147 -0.0120 -0.0111 -0.0082 (-0.6873) (-0.5402) (-0.8471) (-0.6989) (-0.6376) (-0.4769) HHI NFI  -0.1692 -0.1315 (-0.7426) (-0.5787) HHI LFI  0.2072 0.2599 (0.6791) (0.8538) HHI COMB 0.2772 0.2626 (1.0048) (0.9504) Constant 1.40 21 0.4013 1.31 24 0.29 56 1.0776 0.0998 (0.9689) (0.2759) (0.9065) (0.2035) (0.7291) (0.0673) Observations 42 7 427 42 7 42 7 42 7 42 7 Credit Unions 61 61 61 61 61 61 Wald chi2 757.05 767.47 767.40 754.19 767.40 754.19 Prob> chi2 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Hausman chi2(12) 5.56 0.26 4.01 Prob>chi2 0.9365 1.0000 0.9832 Breusch-Pagan (LM)chi2 183 .28 164 .03 184 .60 0.0000 0.0000 0.0000 t-stat istics in parentheses*** p< 0.01, ** p<0.05, * p<0.1 118 University of Ghana http://ugspace.ug.edu.gh Table 4.2 clearly indicates that credit union income from non-loan activities plays a critical role in overall performance; although statistically insignificant, it does have some managerial implication. The results in model 1 and 2 of Table 4.2 suggest that diversification within non-financial income category has coefficient of -0.1692 and -0.1315 though not significant. The negative relation of non-financial income to financial performance affirms the earlier findings of Deyoung and Roland (2001), De joughe (2010) and Florderlisi et al (2011) who hold the view that income diversification affects banking institutions negatively, an observation not different from that of Trinarningsih,Husa, Untoro, Trinugroho and Sutaryo (2016). In direct comparison to the work of Esho et al (2005) study on Australian credit unions, non-financial income is associated negatively with risk adjusted return. Liquid-financial investment measured by HHI  can be said to play a complementing role LFI in financial performance measured by RAROE and RAROA in model 3 and model 4 the coefficient are 0.2072 and 0.2599 in Table 4.2. Since Ghanaian credit unions invest heavily in government treasury bills, the positive association between liquid-financial investment and credit performance is in line with the opinion of Meslier, Tacneng and Tarazi (2014) that non-interest income from government securities affects performance positively. Non-loan income is also associated positively to risk adjusted performance in the case of model 5 and model 6 with positive coefficient 0.2772 and 0.2626. The positive but insignificant relationship compares to conclusions reached by DeYoung and Rice (2004) for 119 University of Ghana http://ugspace.ug.edu.gh the US, Chiorazzo et al (2008) on Italy and Busch and Kick (2009) on German banks. On the other hand, this direct relation between non-loan income and financial performance is in direct contradiction to Deng and Elyasianni (2008) as well as Laeven and Levine (2007) who see diversification as a value decreasing activity. The positive relation between credit union non-loan income in our study compares directly with the conclusion of Goddard, McKillop and Wilson’s 2008 empirical evidence that diversification of credit union income has a positive effect on credit union financial performance. In relation to the non-loan income, the findings of our study with sample Ghanaian credit unions can be positioned with Senya and Wolfe’s (2011) who advanced the argument that diversification certainly impacts performance of financial institutions, particularly banks in developing economies. The size, loan to asset, net interest margin and liquid asset to total asset and Zscore are positive significantly associated with credit union risk adjusted performance hence a percentage increase in these variables leads to a respective increase in credit union risk adjusted performance as presented in Table 4.2. The net worth to total asset and bad loans to loans granted exhibit a negative and significant relationship with credit union risk adjusted return on equity and risk adjusted return on asset. The inference from Table 4.2 is that credit unions from our sample do engage in some non- loan income generating activities. However, these activities are not statistically significant 120 University of Ghana http://ugspace.ug.edu.gh drivers of credit union performance. Rather, they appear slanted towards diversification in the case of non-financial income and are more of a complimentary as an income smoothening strategy that can be adopted by managers. Lastly, loan income is a major and significant driver of credit union financial performance. Objective 2; evaluation of credit union factors on non-loan income. Under this objective we attempt to find out how internal and external factors tend to influence credit union non-loan income hence, what drives non-financial income? The model to support this is specified in equation 3.7. The first discussion is on non-financial income as presented in Table 4.3a and Table 4.3b, followed by discussion of results captured in Table 4.4a and Table 4.4b which is dedicated to diversification within the liquid- financial investment activities of the credit union. The result for this objective from the Tobit estimation is also presented in Table 4.5. The case for combined non-loan income is presented in Table 4.6a and Table 4.6b. Model Diagnostic tests We present empirical results for the non-financial income category of the non-loan income of the credit union specified in equation 3.7. The overall model is significant in explaining the drivers of credit union non-financial income as showed from the Wald chi test scores of 10.7.31, 109.93, 107.09 and 110.00 for the full model 1, the model with parish, workplace and community ownership respectively captured using dummy variables all have probability chi square of 0.000 in Table 4.3a. The Hausman test also confirms the appropriateness of the 121 University of Ghana http://ugspace.ug.edu.gh random effect. The Wald chi test scores of 107.25, 105.59 and 106.90 support our claim that all the parameters used in the model are jointly significant in the Hausman-Taylor model results presented in Table 4.3.b. 122 University of Ghana http://ugspace.ug.edu.gh Table 4.4a: Determinants of Non-Financial Income Random Effects VARIABLES (1) (2) (3) (4) SIZE -0.0905*** -0.1003*** -0.0902*** -0.0953*** (-3.0786) (-3.3339) (-3.0557) (-3.2337) LATA -0.0074*** -0.0072*** -0.0074*** -0.0075*** (-6.5713) (-6.3058) (-6.5388) (-6.6565) L OTA -0.0035*** -0.0034*** -0.0035*** -0.0035*** (-3.0182) (-2.9021) (-3.0269) (-3.0332) N WTA 0.0041** 0.0042** 0.0041** 0.0042** (2.0374) (2.1220) (2.0348) (2.0878) ZSCORE -0.0030 0.0000 -0.0031 -0.0023 (-0.0623) (0.0002) (-0.0652) (-0.0495) AGE -0.0002 -0.0010 0.0000 0.0001 (-0.0807) (-0.4886) (0.0070) (0.0309) COTI -0.0008*** -0.0007** -0.0008*** -0.0008*** (-2.6092) (-2.4776) (-2.6283) (-2.6577) NIM -0.0009 -0.0009 -0.0009 -0.0010 (-0.9805) (-1.0130) (-0.9912) (-1.0519) B KCN 0.0041 0.0031 0.0042 0.0038 (1.1662) (0.8561) (1.1875) (1.0893) INFL -0.0029 -0.0028 -0.0030 -0.0029 (-0.9564) (-0.9015) (-0.9612) (-0.9413) G PD 0.0087** 0.0087** 0.0087** 0.0088** (2.3576) (2.3517) (2.3595) (2.3687) PARI -0.0932 (-1.4978) W ORK -0.0 138 (-0.3237) C OMM 0.07 21 (1.5256) C onstant 0.653 9** 0.741 7** 0.657 5** 0.6668** (2.1038) (2.3473) (2.1105) (2.1501) Observations 427 427 427 427 Credit Unions 61 61 61 61 Wald chi2 107.56 110.08 107.37 110.32 Prob> chi2 0.0000 0.0000 0.0000 0.0000 Hausman chi(11) 10.49 8.56 10.78 9.81 Prob>chi2 0.4870 0.6621 0.4619 0.5471 t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 Table 4.4a.i. Test of difference in Coefficient for type of Credit Union. Type of Credit union chi(1) Prob>chi2 Parish 2.24 0.134 Workplace 0.1 0.7461 Community 2.33 0.1271 123 University of Ghana http://ugspace.ug.edu.gh Table 4.3a indicates that in the random effects estimation, the size of the credit union is significant and inversely connected with non-financial income. The reported coefficients for the parish, workplace and community based credit union are -0.0905, -0.1003, -0.0902 and - 0.0953 respectively. Further the test of significance in difference of the coefficients in Table 4.4a.i show that there is no significant difference in these estimates for the type of credit union. The smaller the size of the credit union, the more it tends to concentrate on one type of income within the non-financial income category of non-loan income, this relationship aligns with that observed by Chiorazzo et al (2008) that small size banks tend to profit more from high level of concentration within income diversification activities, this, also affirms Mercierca et al (2008) who posit that small banks do not benefit from diversification. It might be the case that small sized credit unions in Ghana do not have high demand for loans or the management of smaller sized credit union prefer to engage in more non-financial income generating activities than large size credit union. The negative coefficient identified for loan to asset ratio provides the indication higher levels of loan to asset implies lower levels of diversification within the non-financial activities. This is not particularly shocking as granting of more loans would present the credit union with fewer funds available for other income generating activities. NWTA points to the fact that credit unions with high level of net worth tend to concentrate their income sources within the non-financial income activities. The reason for this is probably that by their nature credit unions have high proportion of equity funds in their capital structure, which serves as a cushion in the face of risk, then propels credit union with higher levels of 124 University of Ghana http://ugspace.ug.edu.gh net worth to diversify less within the non-financial income as the high equity stake can protect them from any negative consequences from their lack of diversification that is, they have the needed cushion in their net worth. Liquid-financial investment to total asset exhibit a significant negative coefficient of 0.0074, in relation to non-financial income, in model (2), model (3) and model (4) the coefficients are 0.0072, 0.0074, 0.0075, when the ownership dummy of parish, workplace and community based ownership are considered. The inference here is that higher investment in liquid-financial investment tends to drain credit union of available funds to engage in non- financial income generating activities which would demand a good spread of income within the non-financial income category of non-loan income. From the results in Table 4a, higher resources usage measure by COTI would mean good spread of income from non-financial activities. The coefficient points to the fact that a 1 percent increase in cost to income leads to a decline of 0.0008 unit of non-financial income if the other variables remain the same, a similar conclusion with coefficient of -0.0007 and -0.0008 is made for the model with ownership dummies included. The solvency indicator, the Z-score is not a significant driver of credit union non-financial income although there is an implication of less solvency exposure to income from non- financial activities on the credit unions sampled for this study. This parallels Templeton and Serveriencse (1992) and Froot and Stein (1998) views that diversification is a hedge against insolvency risk that reduces profitability. Increase in net interest margin from loan activities implies a good spread of income within the non-financial activities but with no significant effect. The age of the credit union is also not significant in our estimation. 125 University of Ghana http://ugspace.ug.edu.gh On the external variables, the growth in the economy measured by GDP is positively associated with non-financial income of the credit union. This observation from Table 4a is intuitive with economic growth theory that as economies grow, business entities within that economy can exploit additional income generating sources. In the case of the sampled credit union, with a growing economy, opportunities to invest funds can be identified easily aside the traditional making of loans, hence credit unions would concentrated on areas within the non-financial income category that gives higher returns. Top 3 bank asset concentration and INFL levels in the economy do little to explain credit union non-financial income. The type of credit union, be it parish, workplace or community based are not significant in the credit union non-financial income process. The Hausman Taylor model results presented in Table 4.3b is used to explain the same non- financial income; in this set of Hausman Taylor estimations the size of the credit union and cost to income is used as endogenous variables. The first set of results is presented in Table 4b. 126 University of Ghana http://ugspace.ug.edu.gh Table: 4.4b Non-Financial Income Hausman Taylor VARIABLES (1) (2) (3) SIZE -0.1152*** -0.1130*** -0.1115*** (-3.0055) (-2.9153) (-2.8693) LATA -0.0070*** -0.0072*** -0.0072*** (-6.0823) (-6.2734) (-6.3619) L OTA -0.0034*** -0.0035*** -0.0035*** (-2.9109) (-3.0000) (-3.0053) N WTA 0.0044** 0.0043** 0.0043** (2.1860) (2.1132) (2.1509) ZSCORE 0.0043 0.0032 0.0028 (0.0776) (0.0568) (0.0512) AGE -0.0006 0.0006 0.0005 (-0.2627) (0.2355) (0.2101) COTI -0.0009*** -0.0009*** -0.0009*** (-2.7567) (-2.7925) (-2.8238) N IM -0.0010 -0.0010 -0.0011 (-1.1354) (-1.1010) (-1.1507) B KCN 0.0024 0.0031 0.0031 (0.6353) (0.8259) (0.8370) I NFL -0.0027 -0.0028 -0.0028 (-0.8909) (-0.9287) (-0.9301) GDP 0.0087** 0.0088** 0.0088** (2.4201) (2.4214) (2.4262) P ARI -0.0995 (-1.3563) W ORK -0.0 112 (-0.2245) COMM 0.07 47 (1.3420) Constant 0.829 9** 0.787 5** 0.7537** (2.3256) (2.2197) (2.1305) O bservations 427 427 427 Credit Union 61 61 61 Wald chi2 107.25 105.59 106.90 Prob> chi2 0.0000 0.0000 0.0000 t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 In the instrumental variable estimation model using the Hausman Taylor estimation accessible in Table 4.3b, the covariates of the independent variables are similar to the Random effect model. The exogenous variable size and cost to income are significant and 127 University of Ghana http://ugspace.ug.edu.gh negatively related to the dependent variable non-financial income; the smaller the size of the credit union the better the spread of income within the non-financial income group is again confirmed and higher resource consumption by the manager of the credit union imply good diversification spread in the non-financial income group. 4.3.3. What drives Liquid-Financial Investment? From Table 4.4a, it is observed that liquid investment to total asset records a positive coefficient of 0.0024, implying that credit unions that are highly liquid also have high concentration of investment activities with the liquid-financial investment activities. 128 University of Ghana http://ugspace.ug.edu.gh Table 4.5a: Liquid-Financial Investment Random Effects VARIABLES (1) (2) (3) (4) SIZE -0.0176 -0.0115 -0.0166 -0.0179 (-0.7864) (-0.5069) (-0.7408) (-0.7934) L ATA 0.0024*** 0.0023*** 0.0023*** 0.0024*** (2.8962) (2.7069) (2.7475) (2.8793) L OTA 0.0002 0.0001 0.0002 0.0002 (0.2561) (0.1700) (0.2056) (0.2487) N WTA 0.0022 0.0021 0.0022 0.0022 (1.4641) (1.3901) (1.4544) (1.4706) Z SCORE 0.0056 0.0036 0.0049 0.0059 (0.1479) (0.0967) (0.1317) (0.1548) A GE 0.0028* 0.0034** 0.0032** 0.0028* (1.8975) (2.1913) (2.0861) (1.8856) C OTI 0.0002 0.0002 0.0002 0.0002 (0.8052) (0.6924) (0.7196) (0.8039) NIM -0.0019*** -0.0019*** -0.0019*** -0.0019*** (-2.7065) (-2.6897) (-2.7343) (-2.6991) BKCN -0.0052** -0.0046* -0.0050* -0.0052** (-2.0020) (-1.7174) (-1.9052) (-2.0017) INFL 0.0050** 0.0049** 0.0050** 0.0050** (2.2139) (2.1629) (2.1959) (2.2148) GDP 0.0059** 0.0059** 0.0059** 0.0059** (2.1529) (2.1611) (2.1615) (2.1533) P ARI 0.0628 (1.2912) WORK -0.0 327 (-0.9764) C OMM 0.00 40 (0.1063) C onstant 0.29 21 0.23 55 0.30 02 0.2924 (1.2510) (0.9928) (1.2853) (1.2489) O bservations 427 427 427 427 Credit Union 61 61 61 61 Wald chi2 44.05 45.83 45.01 43.90 Prob> chi2 0.0000 0.0000 0.0000 0.0000 Hausman chi2(11) 9.06 9.15 9.33 9.01 Prob chi2 0.6162 0.6076 0.5919 0.6213 t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 This relationship is highly significant in all models on Table 4.4a. These results show that the more credit unions invest mobilized funds into interest earning financial investment, the 129 University of Ghana http://ugspace.ug.edu.gh more their diversification into liquid-financial investment. The age of the credit union is a significant driver of liquid-financial investment activities. What can be deduced from this is that as credit unions age, they tend to concentrate more on some particular interest earning liquid-financial investment which might be considered to pose less risk to the credit unions that are aged. The issue of learning from the performance of past liquid-financial investment can also be a reason why older credit union would concentrate on some liquid-financial investment. NIM is negatively related to investment in liquid-financial asset. The negative and significant coefficient of 0.0019 displays a conflict between funds granted as loans hence NIM and funds invested in liquid-financial asset. For the mangers of credit unions these results involve a trade off in how much funds a credit union can allocate for lending purpose and how much funds would be available for investment in liquid-financial asset. The more the credit union concentrates on loans, the more the credit union would also diversify its liquid-investment activities as exhibited by the coefficient of the NIM and diversification within the liquid-financial investment category of credit unions. The size, loan to asset, NWTA, z-score, and cost to income of the credit union are not significantly associated with credit union investment in liquid-financial asset. Ownership type in no other way affects the credit union investment in liquid financial asset. The insignificant nature of the zscore compared to Zhou (2014) who concluded that there is no significant relationship between income diversification and bank risk. 130 University of Ghana http://ugspace.ug.edu.gh In reference to the external variables, top 3 bank asset concentration is inversely related to diversification within the liquid-financial investment. This suggests that credit unions tend to have a good spread of investment within the liquid-financial investment category when banking activities are centered around a few controlling big banks, a situation which can put the credit union in a weaker position where they cannot negotiate for higher returns on their liquid-financial investment which is mostly channeled through the banks. INFL is positively associated with credit union liquid-financial investment from Table 4.4b, when INFL rates increase credit union tends to invest more funds into some particular liquid investment which can easily be liquidated when general price level keep increasing. Increasing GDP means credit union would invest more funds into some liquid investment to take advantage of an overall growth of the economy. 131 University of Ghana http://ugspace.ug.edu.gh Table 4.5b: Liquid- Financial Investment Hausman – Taylor VARIABLES (1) (2) (3) SIZE -0.0163 -0.0194 -0.0201 (-0.6673) (-0.8010) (-0.8200) LATA 0.0022*** 0.0022*** 0.0023*** (2.5870) (2.6177) (2.6992) LOTA 0.0000 0.0000 0.0001 (0.0153) (0.0408) (0.0619) N WTA 0.0028* 0.0028* 0.0028* (1.7981) (1.8021) (1.8148) N IM -0.0016** -0.0016** -0.0016** (-2.2575) (-2.2630) (-2.2567) Z SCORE 0.0125 0.0137 0.0145 (0.2650) (0.2894) (0.3039) A GE 0.0034* 0.0033* 0.0029 (1.7535) (1.6757) (1.5196) COTI 0.0002 0.0002 0.0002 (0.8547) (0.8804) (0.9336) B KCN -0.0045* -0.0048* -0.0050* (-1.6902) (-1.8238) (-1.9092) I NFL 0.0049** 0.0050** 0.0050** (2.2229) (2.2478) (2.2665) G DP 0.0058** 0.0058** 0.0058** (2.2008) (2.2029) (2.1991) PARI 0.0598 (0.9681) WORK -0.0 321 (-0.7453) C OMM 0.00 42 (0.0852) C onstant 0.19 11 0.24 84 0.2393 (0.7667) (1.0106) (0.9735) Observations 427 427 427 Credit Unions 61 61 61 Wald chi2 41.39 41.39 36.60 Prob> chi2 0.0000 0.0000 0.0001 t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 In the instrumental variable model in Table 4.4b, similar conclusions to Table 4.4b emerge. NWTA and NIM are the endogenous variables, the most obvious difference between the covariates in the Random Effects and Hausman-Taylor estimation being that NWTA which is insignificant in the Random Effects is significant in the Hausman-Taylor estimation in 132 University of Ghana http://ugspace.ug.edu.gh explaining investment in liquid-financial investment. Also the significant levels for bank concentration have increased in the Hausman-Taylor model for workplace based credit union compared to the Random Effect model. Table 4.6 Non-Financial Income and Liquid-Financial Investment Tobit VARIABLES HHI NFI  HHILFI  SIZE -0.2019*** -0.0107 (-2.6615) (-0.3251) LATA -0.0175*** 0.0030** (-5.4439) (2.3691) L OTA -0.0066** -0.0003 (-2.4554) (-0.1955) N WTA 0.0076 0.0014 (1.2921) (0.6580) ZSCORE -0.0716 0.0172 (-0.6159) (0.3329) A GE 0.0014 0.0038* (0.3881) (1.8838) COTI -0.0029** 0.0002 (-2.0972) (0.7777) NIM -0.0019 -0.0043** (-0.6067) (-2.3778) B KCN 0.0266* -0.0081** (1.8886) (-2.0024) I NFL -0.0202* 0.0080** (-1.7483) (2.2223) GDP 0.0077 0.0096** (0.8076) (2.1913) C onstant 0.9229 0.3067 Constant (1.0968) (0.8767) Observations 427 427 Credit Union 61 61 Wald chi2 41.39 36.60 Prob> chi2 0.0000 0.0001 t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 In Table 4.5 the values of the index calculated for both the non-financial income and liquid- financial investment ranges between 0 and 1. The Tobit estimation method is employed to provide an explanation for how the independent variables explain HHI and HHI  . NFI LFI 133 University of Ghana http://ugspace.ug.edu.gh In the case of non-financial income, a lower level of 0.45 was used and for that of liquid- financial investment a lower level of 0.462. The results from the estimation for non-financial income show that for every 1 unit increase in the size of the credit union, income from this score decreased by 0.2019 percent. Liquid to total asset is negatively associated with a 0.0175 percent decrease in non-financial income for every 1 unit increase in LATA. A unit increase in LOTA also leads to a 0.0062 percent decline in credit union income from non- financial income. An increase in cost to income means a decline in non-financial income by 0.029 percent. From the external variables top 3 bank concentration tend to increase non- financial income activities and INFL is negatively linked non-financial income. The NWTA, ZSCORE, AGE, NIM and GDP growth tend to have no significant effect in the credit union non-financial income generating route from the Tobit estimation. From the Tobit model on liquid-financial investment in the last column of Table 4.5, liquid asset to total asset is seen to be significant and positively influences diversification of investment in liquid-financial asset by 0.0035 percent per a unit increase in liquid to total asset. Age is also positively and significantly associated with diversification in liquid- financial investment. Net interest margin is negatively associated with liquid-financial investment on the external variables top 3 banks concentration is significant and negatively associated with liquid-financial investment INFL and growth in the economy is significant and positively connected with liquid-financial investment in which a unit increase in these variable tend to increase income generate from liquid-financial investment. 134 University of Ghana http://ugspace.ug.edu.gh 4.3.4 The case of Combined Non-Loan Income Table 4.7a: Combined Non-Loan Income Random Effects VARIABLES (1) (2) (3) (4) SIZE -0.0418* -0.0352 -0.0443* -0.0331 (-1.6567) (-1.3761) (-1.7807) (-1.3561) LOT A -0.0006 -0.0007 -0.0004 -0.0005 (-0.6048) (-0.6969) (-0.4599) (-0.5027) NW TA -0.0023 -0.0025 -0.0023 -0.0025 (-1.4525) (-1.5251) (-1.4331) (-1.5767) ZSCO RE 0.0334 0.0316 0.0348 0.0324 (0.7341) (0.6928) (0.7960) (0.7769) LAT A 0.0009 0.0007 0.0012 0.0012 (1.0042) (0.8149) (1.3402) (1.3453) AG E 0.0040** 0.0048** 0.0030* 0.0035** (2.2221) (2.5451) (1.6823) (2.1424) CO TI 0.0001 0.0000 0.0001 0.0001 (0.2264) (0.1120) (0.3523) (0.2403) NI M -0.0025*** -0.0025*** -0.0025*** -0.0025*** (-3.3729) (-3.3548) (-3.3609) (-3.3279) BK CN -0.0032 -0.0024 -0.0038 -0.0028 (-1.1383) (-0.8350) (-1.3471) (-0.9891) INF L 0.0008 0.0007 0.0009 0.0007 (0.3472) (0.2893) (0.3857) (0.3094) GD P -0.0086*** -0.0086*** -0.0087*** -0.0087*** (-2.9864) (-2.9845) (-2.9965) (-3.0099) PA RI 0.0881 (1.4778) WO RK 0.084 6** (2.1497) COM M -0.159 4*** (-3.7940) Cons tant 1.155 3*** 1.083 7*** 1.133 3*** 1.1354*** (4.4875) (4.1414) (4.4336) (4.5127) Observations 427 427 427 427 Credit Union 61 61 61 61 Wald chi2 47.62 49.90 53.25 64.85 Breusch –Pagan(LM) chi2(01) 319.99 318.12 276.29 246.84 Prob>chibar2 0.0000 0.0000 0.0000 0.0000 t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 Table 4.6a demonstrates that the size of the credit union is significant and inversely associated with income derived from all non-loan activities; this relationship also holds in 135 University of Ghana http://ugspace.ug.edu.gh model (3) which is on workplace based credit union. This relationship is not significant in model (2) and model (4) where parish and community ownership are dummy variables. The inference from the coefficient of size is that increase of credit unions would imply a good spread of income from non-loan income sources. Again, as credit unions age, they tend to show a positive link with total non-loan income; this can be attributed to more knowledge about the market and the need to diversify operations into non-loan activities. Liquid to total asset is positive just as in Meslier, Tacneng and Tarazi (2014), but not a significant driver of combined non-loan income in our case. The net interest margin displays a consistent negative relationship in all models in Table 4.6a, the more credit unions generate income from loan activities, the good the spread they have from non-loan income activities. COTI, a proxy for efficiency and resource usage by management, is positively associated with diversified credit union income although not statistically significant. This relationship disagrees with the work of Nguyen, Nguyen, Nghiem and Nghiem (2016) that diversification of income improves profit efficiency. There exists an inverse relationship between GDP growth and income generated from non- loan activities; this relationship is highly significant. The inference is that as the economy grows and expands, credit unions turn to loan activities rather than non-loan activities. By diversifying their non-loan income sources in an attempt to reduced perceived risk in non- loan activities that may characterize the growth in economy. The ownership dummy variable of workplace and community credit union is significant and positively and negatively associated with combined non-loan income of the credit union respectively. 136 University of Ghana http://ugspace.ug.edu.gh Table 4.7b: Combined Non-Loan Income Hausman –Taylor Tobit Variables (1) (2) (3) (4) SIZE -0.0352 -0.0424* -0.0330 -0.0529* (-1.3549) (-1.6576) (-1.3194) (-1.7995) L OTA -0.0006 -0.0005 -0.0005 0.0003 (-0.6939) (-0.5333) (-0.5452) (0.2707) N WTA -0.0025 -0.0023 -0.0025 -0.0039** (-1.5243) (-1.4416) (-1.5646) (-2.0825) Z SCORE 0.0316 0.0340 0.0322 0.0247 (0.6598) (0.7124) (0.7188) (0.5184) LATA 0.0007 0.0011 0.0011 0.0014 (0.7660) (1.1510) (1.1964) (1.3205) AGE 0.0048** 0.0030 0.0036** 0.0042** (2.4254) (1.5460) (2.0013) (2.1743) COTI 0.0001 0.0001 0.0001 -0.0002 (0.2665) (0.5127) (0.4006) (-0.5353) N IM -0.0021*** -0.0021*** -0.0021*** -0.0059*** (-2.8156) (-2.8192) (-2.7943) (-3.8500) BKCN -0.0023 -0.0036 -0.0027 -0.0031 (-0.7846) (-1.2646) (-0.9474) (-0.9566) I NFL 0.0007 0.0009 0.0007 -0.0005 (0.2814) (0.3793) (0.3078) (-0.1860) G DP -0.0086*** -0.0087*** -0.0087*** -0.0131*** (-3.0148) (-3.0406) (-3.0458) (-3.4897) PARI 0.0889 (1.4159) W ORK 0.08 36* (1.9343) C OMM -0.160 0*** (-3.5239) C onstant 1.070 0*** 1.115 8*** 1.1261*** 1.426 7*** (4.0659) (4.3193) (4.4313) (4.8389) Observations 427 427 427 427 Number of index 61 61 61 61 Wald chi2 45.90 48.03 58.34 47.96 Prob> chi2 0.0000 0.0000 0.0000 0.0000 t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 Table 4.6b shows that in the Hausman Taylor estimation AGE is positive and significantly linked with diversification within the non-loan income category in model (1) and model (3). The NIM is consistently significant and negatively related to credit union non-loan income, 137 University of Ghana http://ugspace.ug.edu.gh again buttressing the point that loan income generation is a preoccupation of credit unions and a heavier contributor to performance. In the instrumental variable estimation for the combined non-loan income, NIM and GDP are significant in determining combined non-loan income diversification with negative coefficient, implying that increase in these variables reduces combined non-loan income of the credit union. The dummy variable on the ownership of the credit union show that workplace and community based ownership are significantly associated with diversification within combined non-loan income group. There exists a positive relation in the case of work based and negative relation for community based credit union ownership and combined non-loan income respectively. In the credit union where ownership is based on an established bond and close knitting with concentrated ownership by way of the common bond, Saghi-Zedek (2016) can be cited to support the observation that workplace owned credit union being significant may be due to the quality of management personnel while community based ownership suffers due to the diseconomies effect and the lack of good human resource to manage the financial resources of the credit union. In the Tobit estimation in the last column of Table 4.6b, a lower level of 0.452 was used. Size of the credit union is inversely and significant at 1 percent increase in size leads to a - 138 University of Ghana http://ugspace.ug.edu.gh 0.0529 percent decline in non-loan income. In addition, a 1 percent increase in net worth of the credit union displays a -0.0039 decline in non-loan income. The Age of the credit union positively influences the combined non-loan income of the credit union in all cases but for parish owned credit union. 4.4 Credit Union Lending Objective 3; analyse the effects of discretional and non-discretional factors on credit union lending. The empirical model to achieve this objective is specified in a general lending equation for the credit union in equations 3.8 and the estimated model in equation 3.9. Table 4.8 is the correlation matrix for the variables used in the lending objective of this study. 139 University of Ghana http://ugspace.ug.edu.gh Table 4.8: Correlation Matrix for Credit Union Lending Variables LNLOTA SIZE SIZESQ ROE LLP NWTA NIETA HHI (COMB) ZSCOR AGE LENDRATE LNLOTA 1.00 SIZE 0.13 1 .00 SIZE SQ 0.12 1.00 1.00 ROE 0.01 0.21 0.20 1.00 LLP (0.12) (0.02) (0.02) (0.32) 1 .00 NWTA 0.06 0.08 0.08 (0.06) 0.06 1.00 NIETA (0.19) (0.15) (0.14) (0.39) 0.62 0.09 1.00 HHI (COMB) (0.13) (0.05) (0.04) 0.03 (0.07) (0.02) (0.25) 1 .00 ZSCOR (0.01) 0.20 0.20 0.01 (0.14) 0.11 (0.13) 0.07 1 .00 AGE 0.14 0.45 0.46 0.16 (0.13) 0.03 (0.31) 0.25 0.17 1.00 LENDRATE (0.30) 0.12 0.12 0.69 0.07 (0.03) 0.18 (0.09) 0.02 0.01 1.00 BKCN 0.12 (0.37) (0.37) (0.12) (0.05) (0.11) (0.11) (0.05) (0.01) (0.17) (0.14) BOCTA 0.11 (0.41) (0.41) (0.11) (0.08) (0.11) (0.13) (0.07) (0.01) (0.18) (0.14) TBILL (0.03) 0.17 0.18 0.01 0.05 0.02 0.07 0.05 0.00 0.07 0.04 GDPG (0.01) (0.08) (0.09) (0.13) 0.06 0.02 (0.01) (0.14) (0.00) (0.05) (0.09) BKCN BOCTA TBILL GDPG BKCN 1.00 BOCTA 0.69 1.0 0 TBILL 0.03 (0.58) 1.00 GDPG 0.14 0.44 (0.61) 1.00 140 University of Ghana http://ugspace.ug.edu.gh Model Diagnostic tests We present empirical results for credit union lending as specified in equation 3.10. Table 4.7 is the results from the fixed effect model. From Table 4.7, the R-square within are 0.1413, 0.1432 and 0.1703 for the internal model, banking sector model and full model 2. These R- square within are small but not strange in panel model estimation compared to time series estimation. Further as this objective seek to explain how discretional and non-discretional factors associates with credit union lending and not a predictive model a low R-square within is acceptable. The Wald chi test scores of 32.85, 41.77 and 60.08 support our claim that all the parameters used in the model are jointly significant in the model presented in Table 4.7. The Wald chi test scores support our claim that all the parameters used in the model are jointly significant. Their respective probability values are 0.0000 in all estimations. The Wald chi test score of 53.17 in the random effect model is significantly different from zero. From the Hausman test, in the internal model, the banking sector model the fixed effect model is favored. The full model 1 in column 3 the random effects is preferred, whiles in the full model 2, the fixed effected model is proffered. 4.4.1 Lending Empirical Results Discussion In the models estimated in Table 4.7, the internal model uses only variables under the control of the management of the credit unions. The banking sector model, is addition of banking sector variable to the internal model as credit union activities is heavily influenced 141 University of Ghana http://ugspace.ug.edu.gh banking sector activities. The full model 1 consider macroeconomic variables as literature suggest that the macro economy also is important in lending. The last column of Table 4.7 full model 2, assesses the non-linearity of credit union size in the lending relationship. From the results in Table 4.7, size of the credit union is significant and positively associated with the loan to asset ratio of the credit union in the full model, and loan to asset cannot be considered purely with management discretional variables. Under these conditions, the intuitive and logical inference is that the larger the credit union, the more loans it tends to give out. As Smirlock (1985) suggests, credit unions with more substantial total asset tend to grant more loans probably to justify the increase in resources. 142 University of Ghana http://ugspace.ug.edu.gh Table 4.9: Lending Empirical Estimation VARIABLES Internal Model Banking Sector Full Model 1 Full Model 2 SIZE 0.0205 0.0281 0.0439** 0.4518** (0.8181) (0.9771) (2.021) (2.2066) SIZE2SQU -0.0326* (-1.9477) L .ROE -0.0007 -0.0006 -0.0 013 -0.0009 (-0.5856) (-0.4853) (-1.0657) (-0.7036) L LP -0.0041*** -0.0043*** -0.0027* -0.0044*** (-2.7374) (-2.8153) (-1.8981) (-2.8523) N WTA -0.0033** -0.0033* 0.001 -0.0029* (-1.9754) (-1.9585) -0.8177 (-1.7616) N IETA 0.0055** 0.0060** 0.0024 0.0068** (2.1527) (2.2436) (1.0134) (2.5532) HHI COMB -0.0205 -0.0193 -0.0672* -0.0372 (-0.5694) (-0.5308) (-1.9252) (-1.0018) ZSCORE 0.0059*** 0.0060*** 0.0001 0.0059*** (2.994) (3.0242) (0.1443) (3.0162) AGE -0.0096*** -0.0083** 0.0014 -0.0056 (-2.9515) (-2.1182) (1.0529) (-1.3967) LDRT -0.0020*** -0.0020*** -0.0021*** -0.0021*** (-4.2323) (-4.0632) (-4.4943) (-4.2624) B KCN 0.0016 0.0051** 0.0051* (0.7981) (1.9613) (1.8455) B OCTA 0.0014 0.0066 0.0006 (0.1792) (0.6817) (0.0547) TBILL -0.0040** -0.0038* (-2.0250) (-1.9029) G DP -0.0059*** -0.0052*** (-3.2162) (-2.5939) Constant 1.7946*** 1.6552*** 1.3341*** 0.2321 (11.1831) (6.1455) (6.3807) (0.3436) Observations 366 366 366 366 Wald Chi2 32.85 41.77 53.17 60.08 Hausman chi 108.04 45.26 28.2 33.02 Prob>chi2 0.0000 0.0000 0.0085 0.0029 Breusch Pagag LM 247.21 Prob > chibar2 0.0000 R-squared Within 0.1413 0.1432 0.1703 Decision FE FE R E FE t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 FE=Fixed Effects and RE = Random Effects 143 University of Ghana http://ugspace.ug.edu.gh Previous years’ returns on equity also negatively associates with lending in credit unions, a result that is particular surprising as loans make up about 70 percent of credit unions’ income composition in all four models estimated. Loan loss provision is negatively related to credit union loans, implying that increases in bad loans reduces the loan granted by the credit union; this relationship is true in the entire models estimated in Table 4.7. As Quagliariello (2007) also illustrated, provision for loan loss to loans granted negatively associates with credit union lending, and increase in loan loss reduce the amount of loans that can be extended by the credit union. Credit union managers would have to work on reducing bad loans if they want to increase their performing loan portfolio. The quality management measured by NEITA is positively associated with credit union loans; high quality management influences the level of loans that a credit union can create out of deposits. Meanwhile, an increase in the NWTA, of the credit union negatively affects its loans business. In other words, the more solvent a credit union is, the less likely it is to give out loans, a relationship exhibited to be highly significant in Table 4.7 in all models but for the random effect model in column 3. The number of years that the credit union has been in existence positive in the random effect model however negative in the full model in panel 4 although not significant. Attempts by managers of credit unions to add on to their loan income by diversifying into non-loan income activities reduces their lending business as presented in all models of Table 4.7. This is to be expected because, with the limited pool of funds available to credit unions, 144 University of Ghana http://ugspace.ug.edu.gh any attempt to pursue non-loan income activities would lead to reduction in funds available for lending purposes. Management is therefore advised to be cautious in their pursuit of non- loan income due to its risk and competition with funds for loan business. An increase in credit union lending rates has the ability to reduce loan granted to borrowers of the credit union. For instance, a 1 percent increase in lending rate would decrease loans by 0.0020, 0.0020, 0.0021 and 0.0021 percent in all four models respectively in Table 4.7. One reason for this correlation is that if credit union managers decide to increase lending rates, they stand a high chance of losing loan business to other competing financial institutions such as the commercial banks, a situation clearly argued theoretically in Smith, (1984) and Rubin, Overstreet, Beling & Rajaratnam, (2013). With the introduction of non-discretional variables such as financial development variables into our model, the coefficient of size is positive and significant statistically. Top 3 bank asset concentration positively influences credit union loans and significant in the full model 1 and full model 2, with coefficient of 0.0051 at 10 percent significant level and 0.0051 at 5 percent level. The inference here is that as the banking sector gets dominated by a few banks with a high probability of the industry not being competitive, credit unions tend to make more loans, probably because these banks would be charging high rates on loans thereby unintentionally directing loan business to credit unions, a conclusion which agrees with that of Emmons & Schmid (2000), Feinberg (2001) and Ely & Robinson (2009). Thus a non- 145 University of Ghana http://ugspace.ug.edu.gh competitive banking sector with few banks dominating banking activities indirectly creates the room for credit union loan business to thrive. Efficiency of the banking sector measures by bank over cost to total asset also displays a positive but not significant association with credit unions’ loan business, implying that an inefficient banking industry would probably result in overcharging interest on loans, giving credit unions the opportunity to grant more loans. Drawing a similar conclusion, Feinberg (2001) contends that increased credit union activities in a particular market have the ability to discipline banks in the exercise of their market power. Further in model 3 of Table 4.7, macroeconomic variables are added to the model, the one year Treasury bill note rate use as a proxy for monetary policy. The model displays a significant negative relation with credit union loans in the full models 1 and 2; an increase in the one year Treasury bill rate tends to indirectly impact credit unions’ loan business, a result that is not surprising as the proposition here is that an increasing Treasury bill rate implies increasing interest on the market. A plausible reason for this can be that the mangers of the credit union may be tempted to pursue liquid-financial investment which is risk free instead of making more loans. Still, this relation is rather counter intuitive as one expects that during period of contractionary monetary policy, cost of borrowing increase on the loanable funds market. The members of the credit union are by this expected to contract more loans from the credit unions and not from the banks, since loan rate and lower than loans contracted from banks. 146 University of Ghana http://ugspace.ug.edu.gh Growth in the economy does not translate into increased loan business for the credit union as presented in the full models 1 and 2 of Table 4.7. This might be due to the fact that credit unions are member centered in operation, a situation that restricts their participation in the general growth in the economy beyond what the members of the union pursue through the loans that they contract from the credit union, a view that needs further investigation. The full model in panel 2 of Table 4.7 verifies that if there exists a non-linear relationship of size in the credit union loan business and the results show a negative coefficient, the non- linearity relationship is confirmed. The inference is that the size of the credit union positively influences loan business up to a point, after which any increase in size negatively associates with the loan business size of the credit union. The managerial implication is that growing the size of the credit union is important, however managers must tread cautiously so as not to experience any diseconomies of scale in their loan business. 4.5 Efficiency Objective 4; estimate Cost efficiency and Technical efficiency levels credit unions. From Table 4.8, loan granted over the 7 year sampled period was GH₵920,769.50 on the average with a maximum loan of GH₵22,500,000. Liquid investment ranged between GH₵832.34 and GH₵7,560,580. Non-financial income had a variability of GH₵38,903.09, 147 University of Ghana http://ugspace.ug.edu.gh during the sample period some credit unions did not generate Non- financial income, the highest output of non- financial income is GH₵324,865.10. Table 4.10. Descriptive states for variables in First Stage DEA Variable Mean Std. Dev. Min Max Outputs Loan 920,769.50 1,789,777.00 300 2 2,500,000.00 Liquid Investment 444,078.30 898,365.10 832.34 7,560,580.00 Non-Financial Income 17,826.51 38,903.09 0 324,865.10 Inputs Personnel cost 35,407.84 63,221.41 60 469,839.60 Non- Current Asset 77,633.70 147,144.50 10 980,877.40 Regular Savings 1,280,355.00 2,307,906.00 2,874.50 25,200,000.00 Inputs Price Price of Labor 2.6 1.97 0.1 16.27 Price of Non-Current Asset 190.65 3,047.46 0.03 65,331.40 Price of Regular Savings 15.45 5.66 3.16 45.49 The input section of Table 4.8 show that personal cost hovered between GH₵60 to GH₵ 469,839.60. Non-current asset has a mean of GH₵77,633.70. Regular savings ranged between GH₵2,874.50 to GH₵25,200,000. Price of labour averaged GH₵2.57 and a maximum value of GH₵16.27. Price of Non-current asset ranged between GH₵0.03 to GH₵65,331.40. The mean price of regular savings was GH₵15.45. 4.5.1 Cost Efficiency Analysis From Table 4.8a, we measure overall improvement in cost efficiency over the 7 years using the geometric mean (Gavg) because it is less influenced by very small or large values in skewed data. From Table 4.1, the average cost efficiency score from the Tone’s measure 39 percent with a standard deviation of 21.50 percent, the range of scores for cost efficiency for the sample period is 10.78 percent to 100 percent. A close study of the cost efficiency scores 148 University of Ghana http://ugspace.ug.edu.gh overall shows that few credit unions recorded more than 50 percent cost efficiency over the sample period. For brevity we analyse the performance of the top 5 and worst 5 credit union as presented in italics of Table 4.8a. The top 5 most efficient credit unions during the period 2008 to 2014 were Abosom, UG, Dunkwa Trd, KAMCCU and Ebenezer using the arithmetic mean scores of 83.8 percent, 81.3 percent, 73.6 percent, 72.9 percent and 61.4 percent respectively. However ranking these credit unions using geometric mean, Abosom scored 80.9 percent, UG 78.9 percent, KAMCCU 71.9 percent, Dunkwa Trd 67.0 percent and Ebenezer 59.3 percent respectively as top performers. The worst performers by way of cost efficiency were N Fadama 23.7 percent, UGARS 21.5 percent, UniEduwin 20.2 percent, NAFTI 18.6 percent and st. Maggi 13.7 percent. By way of geometric mean, these credit unions scored 22.7 percent, 20.3 percent, 20 percent, 18.5 percent and 13.6 percent respectively. On the whole, cost efficiency has ranged between 33 percent and 44 percent, with an average score of 38.9 percent and with a geometric mean of 33.7 percent. The year 2013 recorded the highest variability at 25 percent, increasing marginally to 25.3 percent in 2014. From this, we conclude that cost efficiency for credit unions is generally very low, implying production costs for most credit unions are very high. 149 University of Ghana http://ugspace.ug.edu.gh Table 4.11a: Cost Efficiency Scores DMU 2008 2009 2010 2011 2012 2013 2014 Avg Gavg Rank ABOSOM 1 0.501 0.573 0.793 1 1 1 0.838 0.809 1 UG 0.489 0.645 0.709 0.941 0.907 1 1 0.813 0.789 2 KAMCCU 0.547 0.577 0.739 0.792 0.843 0.711 0.897 0.729 0.719 3 DUNKWATRD 0.304 0.375 0.628 0.843 1 1 1 0.736 0.670 4 EBENEZER 0.372 0.431 0.556 0.696 0.659 0.806 0.778 0.614 0.593 5 ASAWINSO 0.576 0.465 0.594 1 0.565 0.485 0.586 0.61 0.592 6 DUNAREATEA 0.299 0.281 0.366 0.676 1 1 1 0.66 0.575 7 MINESCHO 0.4 0.284 0.444 0.906 0.903 0.759 0.258 0.565 0.502 8 NKORANATEA 0.233 0.26 0.325 0.424 0.906 1 0.78 0.561 0.480 9 ADOAGYIRI 0.412 0.286 0.392 0.532 0.527 0.585 0.66 0.485 0.469 10 NAVRONGO 0.526 0.485 0.611 0.555 0.375 0.433 0.343 0.475 0.467 11 BAWKU HOS 0.345 0.293 0.423 0.234 0.538 0.674 0.749 0.465 0.430 12 WATER 0.115 0.359 0.47 1 0.42 1 0.316 0.526 0.427 13 STANDARD 0.392 0.306 0.394 0.494 0.674 0.402 0.406 0.438 0.427 13 GH STATS 0.476 0.355 0.583 0.156 1 0.561 0.264 0.485 0.419 14 WEST MANT 0.276 0.568 0.641 0.291 0.626 0.335 0.33 0.438 0.412 15 KEKEKRACHI 0.541 0.36 0.547 0.483 0.234 0.27 0.491 0.418 0.399 16 WESTPOWER 0.29 0.339 0.295 0.318 0.314 0.467 1 0.432 0.389 17 BAW TEACH 0.176 0.222 0.357 0.488 0.694 0.558 0.514 0.43 0.389 17 NKORAN VIC 0.307 0.341 0.357 0.252 0.17 0.969 0.807 0.458 0.385 18 GREL 0.311 0.22 0.223 0.26 0.492 0.681 0.823 0.43 0.378 19 WENCHI 0.299 0.433 0.347 0.403 0.366 0.547 0.231 0.375 0.363 20 CRIG TAFO 0.335 0.257 0.261 0.332 0.367 0.49 0.592 0.376 0.361 21 TEC ARE TEA 0.196 0.226 0.227 0.34 0.555 0.574 0.583 0.386 0.349 22 SEGE 0.328 0.298 0.368 0.614 0.443 0.258 0.219 0.361 0.342 23 GRA 0.232 0.282 0.234 0.347 0.364 0.562 0.45 0.353 0.337 24 JACCU 0.228 0.24 0.32 0.325 0.116 0.823 0.828 0.411 0.333 25 UG MED 0.394 0.288 0.386 0.343 0.346 0.348 0.241 0.335 0.331 26 KWAEBIBI 0.3 0.274 0.242 0.416 0.335 0.234 0.557 0.337 0.322 27 TRINITY 0.538 0.394 0.334 0.293 0.25 0.214 0.319 0.335 0.321 28 BOLE CATH 0.399 0.392 0.41 0.305 0.243 0.113 0.606 0.353 0.318 29 ANAJICHRIST 0.222 0.195 0.414 0.395 0.418 0.316 0.351 0.33 0.318 29 ECG 0.479 0.265 0.293 0.267 0.268 0.284 0.428 0.326 0.317 30 WAWORKERS 0.289 0.239 0.307 0.277 0.334 0.309 0.518 0.325 0.316 31 CHR OF PENT 0.256 0.294 0.305 0.305 0.275 0.393 0.389 0.317 0.313 32 TRINITYGAR 0.327 0.28 0.321 0.172 0.34 0.37 0.374 0.312 0.304 33 SOIL RESEARCH 0.321 0.181 0.204 0.291 0.365 0.417 0.428 0.315 0.301 34 TUC NAT 0.217 0.248 0.315 0.289 0.287 0.375 0.413 0.306 0.300 35 ACC ACA 0.783 0.503 0.156 0.287 0.201 0.207 0.293 0.347 0.299 36 APAPAM 0.253 0.36 0.403 0.347 0.192 0.265 0.253 0.296 0.288 37 150 University of Ghana http://ugspace.ug.edu.gh AAK TEACH 0.34 0.316 0.234 0.299 0.203 0.28 0.338 0.287 0.283 38 STPAUL 0.338 0.27 0.314 0.295 0.282 0.219 0.278 0.285 0.283 38 UDS 0.295 0.301 0.275 0.31 0.317 0.165 0.342 0.286 0.280 39 ATICO 0.509 0.414 0.209 0.26 0.278 0.243 0.171 0.298 0.279 40 STJOSEPH 0.258 0.204 0.257 0.227 0.316 0.305 0.416 0.283 0.276 41 KADJEBI TEA 0.373 0.361 0.367 0.227 0.249 0.188 0.227 0.285 0.275 42 GHANA NAT CUL 0.223 0.218 0.23 0.246 0.348 0.373 0.322 0.28 0.274 43 GUTA 0.353 0.305 0.392 0.328 0.148 0.226 0.246 0.285 0.273 44 KPANDNEWERA 0.215 0.289 0.304 0.305 0.202 0.315 0.175 0.258 0.252 45 ADIDO TEA 0.193 0.272 0.298 0.258 0.359 0.279 0.137 0.257 0.247 46 BUNSOCRIG 0.285 0.278 0.267 0.235 0.233 0.234 0.206 0.248 0.247 46 NKAWKAW 0.288 0.199 0.203 0.317 0.216 0.223 0.276 0.246 0.242 47 JAS/ GOLDEN 0.2 0.226 0.324 0.21 0.371 0.193 0.177 0.243 0.234 48 ALU 0.248 0.23 0.249 0.252 0.232 0.209 0.224 0.235 0.234 48 SAMATEX 0.22 0.182 0.179 0.197 0.244 0.319 0.32 0.237 0.231 49 GHATOMIC 0.255 0.13 0.223 0.369 0.254 0.195 0.231 0.237 0.227 50 N FADAMA 0.323 0.265 0.275 0.188 0.212 0.27 0.123 0.237 0.227 50 UGARS 0.134 0.131 0.181 0.23 0.222 0.231 0.377 0.215 0.203 51 UNIEDU WIN 0.226 0.234 0.197 0.183 0.171 0.174 0.228 0.202 0.200 52 NAFTI 0.212 0.204 0.182 0.142 0.19 0.175 0.196 0.186 0.185 53 ST MAGGI 0.108 0.178 0.136 0.135 0.143 0.138 0.122 0.137 0.136 54 MEAN 0.3 34 0.3 08 0.3 5 0.3 88 0.4 12 0.4 39 0.4 46 0.3 82 0.3 60 GAVG 0.307 0.292 0.325 0.342 0.352 0.371 0.381 0.356 0.337 STD. DEV. 0.152 0.108 0.143 0.221 0.249 0.265 0.257 0.154 0.141 4.5.2 Technical Efficiency Analysis From Table 4.8b the average technical efficiency score from the input VRS is 54.39 percent with a standard deviation of 25.14 percent, the range of scores for technical efficiency for the sample period is 11.02 percent to 100 percent. Similarly for brevity we analyse the performance of the top 5 and worst 5 technically efficient credit union as presented in italics of Table 4.8b. The results in Table 4.8b, is from the VRS technical efficiency method, the most technically efficient credit union over the 151 University of Ghana http://ugspace.ug.edu.gh period was the GRA at 97.1 percent followed by Asawinso 95.4 percent, Minechso 93 percent, KAMCCU 90.4 percent and TUC Nat 89.8percent. Their geometric mean is 97 percent, 94.9 percent, 90.8 percent, 89.8 percent and 88.1 percent respectively. The credit unions with the least technical efficiency are GUTA, Samatex, UG Med Kadjebi Tea and UniEduwin with technical efficiency scores of 28.6 percent, 25.5 percent, 24.4 percent, 25 percent and 21.1 percent respectively. The geometric mean for these credit unions is 28.6 percent, 25.3 percent, 24.4 percent, 24.2 percent and 20.9 percent respectively. On the whole, the average technical efficiency ranged between 50.5 percent to 59.7 percent with an average of 54.4 percent and a geometric mean in technical efficiency of 47.9 percent. There has been more variation in technical efficiency as seen in the standard deviation ranging from 21.7 percent to 27.4 percent with an average of 19.4 percent in Table 4.1. We observe from Table 4.8b that resources are being used more efficiently in credit unions thereby avoiding a lot of waste compared to the production process with low cost efficiency scores. 152 University of Ghana http://ugspace.ug.edu.gh Table 4.11b: Technical Efficiency Scores DMU 2008 2009 2010 2011 2012 2013 2014 Avg Gavg Rank GRA 1 0.979 1 1 0.926 1 0.892 0.971 0.970 1 ASAWINSO 1 0.949 1 1 1 1 0.732 0.954 0.949 2 MINESCHO 1 1 1 1 1 1 0.509 0.93 0.908 3 KAMCCU 0.737 0.811 0.97 1 1 0.81 1 0.904 0.898 4 TUC NAT 0.602 0.695 1 0.986 1 1 1 0.898 0.881 5 UG 0.547 0.639 0.888 1 1 1 1 0.868 0.846 6 ABOSOM 1 0.526 0.65 0.765 1 1 1 0.849 0.826 7 WATER 0.958 0.698 0.641 0.675 0.892 1 0.774 0.806 0.794 8 EBENEZER 0.449 0.54 0.681 0.889 0.804 0.887 1 0.75 0.724 9 GHANA NAT CUL 1 0.82 0.625 0.521 0.618 0.636 0.78 0.714 0.699 10 DUNKWATRD 0.357 0.4 0.569 0.732 0.91 1 1 0.71 0.659 11 STPAUL 0.938 0.772 0.788 1 0.512 0.314 0.429 0.679 0.630 12 ACC ACA 1 0.662 0.427 0.793 0.67 0.509 0.503 0.652 0.628 13 ECG 1 1 0.363 0.385 0.548 0.535 0.805 0.662 0.614 14 NKORANATEA 0.384 0.386 0.438 0.553 0.916 1 0.983 0.666 0.612 15 GH STATS 1 0.462 0.62 0.373 1 0.936 0.309 0.672 0.609 16 WENCHI 1 1 0.978 0.558 0.383 0.36 0.382 0.666 0.602 17 ADOAGYIRI 0.388 0.393 0.553 0.634 0.709 0.717 0.809 0.6 0.580 18 BAWKU HOS 0.548 0.365 0.642 0.487 0.585 0.723 0.826 0.597 0.579 19 NAVRONGO 0.664 0.584 0.616 0.52 0.532 0.63 0.485 0.576 0.573 20 N FADAMA 0.515 0.604 0.356 0.455 0.552 0.656 0.956 0.585 0.561 21 NAFTI 0.414 0.564 0.595 0.762 0.589 0.53 0.462 0.559 0.550 22 TRINITY 0.727 0.566 0.611 0.591 0.424 0.39 0.499 0.544 0.533 23 STANDARD 0.526 0.379 0.451 0.571 0.636 0.617 0.526 0.529 0.522 24 ATICO 1 1 0.392 0.418 0.488 0.389 0.319 0.572 0.517 25 JAS/ GOLDEN 0.281 0.313 0.536 0.384 0.684 0.936 0.831 0.567 0.515 26 ANAJICHRIST 0.258 0.276 0.58 0.665 0.745 0.649 0.709 0.555 0.514 27 WEST POWER 0.473 0.505 0.448 0.405 0.375 0.582 1 0.541 0.514 27 APAPAM 0.597 0.454 0.484 0.503 0.476 0.525 0.524 0.509 0.507 28 WEST MANT 0.678 0.701 0.624 0.372 0.429 0.372 0.402 0.511 0.493 29 BAW TEACH 0.301 0.316 0.394 0.608 0.718 0.569 0.647 0.508 0.482 30 CRIG TAFO 0.42 0.589 0.391 0.451 0.401 0.471 0.694 0.488 0.478 31 JACCU 0.34 0.348 0.343 0.388 0.319 1 1 0.534 0.469 32 DUNAREATEA 0.202 0.249 0.31 0.439 0.783 0.893 0.944 0.546 0.462 33 STJOSEPH 0.56 0.392 0.412 0.357 0.482 0.518 0.535 0.465 0.459 34 ST MAGGI 0.502 0.51 0.42 0.401 0.315 0.376 0.728 0.465 0.450 35 ADIDO TEA 0.379 0.491 0.504 0.591 0.606 0.501 0.178 0.464 0.436 36 BOLE CATH 0.507 0.432 0.421 0.466 0.396 0.35 0.476 0.435 0.433 37 WAWORKERS 0.341 0.351 0.423 0.359 0.473 0.426 0.753 0.447 0.431 38 CHR OF PENT 0.351 0.397 0.433 0.373 0.385 0.546 0.514 0.428 0.423 39 153 University of Ghana http://ugspace.ug.edu.gh SEGE 0.359 0.434 0.493 0.56 0.628 0.326 0.254 0.436 0.418 40 TEC ARE TEA 0.293 0.302 0.326 0.366 0.558 0.589 0.629 0.438 0.417 41 GHATOMIC 0.326 0.269 0.353 0.442 0.535 0.535 0.454 0.416 0.405 42 SOIL RESEARCH 0.471 0.278 0.268 0.33 0.383 0.497 0.533 0.394 0.381 43 TRINITYGAR 0.785 0.43 0.258 0.334 0.306 0.302 0.377 0.399 0.373 44 UDS 0.327 0.38 0.391 0.435 0.419 0.312 0.364 0.375 0.373 44 GREL 0.247 0.246 0.282 0.268 0.441 0.604 0.743 0.404 0.368 45 UGARS 0.332 0.316 0.406 0.348 0.313 0.322 0.516 0.365 0.359 46 NKAWKAW 0.316 0.327 0.301 0.349 0.381 0.373 0.477 0.361 0.357 47 KWAEBIBI 0.309 0.291 0.285 0.458 0.353 0.301 0.56 0.365 0.354 48 BUNSOCRIG 0.426 0.367 0.338 0.345 0.362 0.312 0.303 0.35 0.348 49 KPANDNEWERA 0.328 0.429 0.568 0.448 0.206 0.262 0.229 0.353 0.332 50 NKORAN VIC 0.391 0.389 0.417 0.123 0.11 0.709 0.71 0.407 0.331 51 ALU 0.411 0.262 0.253 0.306 0.305 0.272 0.232 0.292 0.287 52 AAK TEACH 0.243 0.306 0.297 0.246 0.243 0.293 0.374 0.286 0.283 53 KEKEKRACHI 0.401 0.275 0.315 0.224 0.215 0.256 0.332 0.288 0.282 54 GUTA 0.405 0.291 0.321 0.281 0.249 0.223 0.235 0.286 0.281 55 SAMATEX 0.233 0.226 0.252 0.266 0.226 0.256 0.329 0.255 0.253 56 UG MED 0.259 0.234 0.261 0.239 0.237 0.254 0.226 0.244 0.244 57 KADJEBI TEA 0.34 0.298 0.304 0.201 0.195 0.161 0.249 0.25 0.242 58 UNIEDU WIN 0.22 0.213 0.172 0.189 0.199 0.2 0.287 0.211 0.209 59 Mean 0.5 31 0.4 87 0.4 99 0.5 11 0.5 43 0.5 69 0.5 96 0.5 34 0.5 12 Gavg 0.474 0.442 0.458 0.461 0.483 0.507 0.536 0.5 0.479 Std. Dev. 0.264 0.225 0.218 0.236 0.255 0.266 0.258 0.193 0.189 4.5.3 Comparing Cost Efficiency and Technical Efficiency. Objective 5; discuss the interrelation present between cost efficiency and technical efficiency on the sample credit union. From Figure 4.1, using trend, the improvement in Tone’s efficiency growth was higher than technical efficiency for the period 2008 to 2011, from 2012 to 2014, there was much improvement in the sampled credit union technical efficiency than its Tone’s efficiency. 154 University of Ghana http://ugspace.ug.edu.gh Figure 4.1: Growth Trend in Tones Efficiency and Technical Efficiency From Table 4.12 Panel A, the test of difference in the series for Tones efficiency and Technical efficiency using the non-parametric Wilcoxon Singed–Rank test show that there exist a significant difference in the Tones efficiency and Technical efficiency for credit unions. 155 University of Ghana http://ugspace.ug.edu.gh Table 4.12. Non-Parametric Wilcoxon Singed-Rank Test Panel A: Year Positive Negative Zero Z-statistics Prob > Z 2008 1752 138 1 5.797 0.000 2009 1820.5 70.5 0 6.285 0.000 2010 1724 167 0 5.592 0.000 2011 1594.5 295.5 1 4.665 0.000 2012 1642 246 3 5.014 0.000 2013 1615 261 15 4.864 0.000 2014 1722.5 158.5 10 5.618 0.000 Panel B: Growth in Efficiency Wilcoxon Singed-Rank Test Positive Negative Zero Z-statistics Prob > Z 2008 to 2014 19 8 1 0.933 0.3508 In Panel B, the Wilcoxon Singed-Rank test on growth in the two efficiency measures, show that there is no difference in the growth of Tone’s efficiency and Technical efficiency for the period 2008 to 2014. For brevity, we analyse the performance of the top 3 credit unions, comparing their cost and technical efficiency scores, with the bottom 3 credit unions’ technical and cost efficiency score in Table 4.8a and 4.8b, to verify if technical efficiency translates into cost technical efficiency or vice versa. Abosom ranked 1st with a score of 80.9 percent cost efficiency but was ranked 7th in technical efficiency with an even higher score of 82.6 percent. Similarly, the second ranked cost efficient credit union, UG with a score of 78.9 percent is ranked 6th under technical efficiency with a score of 84.6 percent a higher score than cost efficiency. The 3rd ranked cost efficient credit union, KAMCCU, with a score of 71.9 percent, ranked 4th under technical efficiency with score of 89.8 percent. 156 University of Ghana http://ugspace.ug.edu.gh The worst ranked technical efficient credit union, Uni of Edu, with a score of 20.9 percent ranked 59th, but ranked 52nd under cost efficiency with a score of 20 percent. The next worst ranked credit union under technical efficiency is Kadjebi Tea scored 24.2 percent ranked 58th but ranked 42nd under cost efficiency with a score of 27.5 percent. Finally, the 3rd from bottom ranked based on technical efficiency, UG Med, scored 24.4 percent, ranked 57th but took the 26th position in cost efficiency with a score of 33.3 percent. We can conclude from this evidence in Table 4.8a and 4.8b that technical efficiency does not necessarily translate into cost efficiency and vice versa. The reasons and implications of this are varied and the current work does not seek to answer this. From Table 4.13, the study concludes from the paired T-test conducted on the Tone’s efficiency and Technical efficiency that significant difference exist between the Tone efficiency and Technical efficient, hence the two series is significantly different. Table 4.13. Paired T- test Year Mean Difference t-statistics Degree of Freedom 2008 .1965 6.5780 60 2009 .1782 6.8819 60 2010 .1487 6.1258 60 2011 .1228 4.7376 60 2012 .1319 5.8008 60 2013 .1305 5.5224 60 2014 .1495 6.0357 60 157 University of Ghana http://ugspace.ug.edu.gh 4.6 Determinants of Credit Union Cost efficiency and Technical efficiency Objective 7; empirically examine the determinants of credit union cost efficiency and technical efficiency. Table 4.14a and Table 4.14b is the correlation matrix for the variables used in the efficiency objective of the study. 158 University of Ghana http://ugspace.ug.edu.gh Table 4.14a: Correlation Matrix for Cost Effieicny and Technnical Efficiency Variables CE TE RAROE RAROA SIZE ZSCOR NWTA LLP HHI (NFI ) HHI HHI (LFI ) (COMB) NIETA CE 1.00  0.64 1.0 0 TE RAROE 0.30 0.17 1.0 0 RAROA 0.30 0.17 1.00 1.0 0 SIZE 0.51 0.13 0.35 0.35 1.0 0 ZSCOR 0.07 (0.01) 0.72 0.73 0.21 1.0 0 NWTA 0.10 (0.08) 0.05 0.08 0.08 0.11 1.0 0 LLP (0.21) (0.08) (0.29) (0.28) (0.03) (0.14) 0.06 1.0 0 HHI (NFI ) 0.01 (0.20) (0.12) (0.12) (0.13) (0.00) (0.02) (0.07) 1.0 0 HHI (LFI ) 0.20 0.25 0.00 0.01 0.01 (0.01) 0.05 0.08 (0.18) 1.0 0 HHI (COMB) 0.02 0.28 0.11 0.10 (0.04) 0.07 (0.03) (0.08) (0.41) 0.38 1.0 0 NIETA (0.30) (0.12) (0.38) (0.37) (0.14) (0.13) 0.08 0.60 0.02 (0.07) (0.25) 1.0 0 LOTA (0.12) (0.28) (0.06) (0.06) 0.09 0.01 0.08 (0.13) 0.25 (0.10) (0.12) (0.20) AGE 0.36 0.18 0.26 0.26 0.45 0.17 0.03 (0.14) (0.09) 0.16 0.25 (0.31) LITA 0.25 0.47 0.07 0.08 (0.12) (0.07) 0.07 0.14 (0.41) 0.22 0.20 0.17 BKCN (0.18) (0.08) (0.06) (0.07) (0.37) (0.01) (0.11) (0.05) 0.16 (0.07) (0.05) (0.11) BKZSCOR 0.21 0.10 0.02 0.02 0.43 0.01 0.12 0.13 (0.17) 0.07 0.04 0.14 BOCTA (0.20) (0.13) (0.02) (0.03) (0.41) (0.00) (0.11) (0.08) 0.20 (0.08) (0.07) (0.13) COTI (0.35) (0.18) (0.59) (0.58) (0.24) (0.13) (0.00) 0.44 0.04 0.01 (0.09) 0.64 NIM (0.11) (0.06) 0.18 0.18 0.08 (0.03) (0.01) 0.10 (0.04) (0.18) (0.21) 0.25 RTBILL 0.08 0.03 (0.07) (0.06) 0.18 0.00 0.06 0.10 0.02 0.03 (0.04) 0.08 GPDG (0.01) (0.05) (0.07) (0.06) (0.09) 0.00 0.01 0.07 0.16 0.03 (0.14) (0.01) 159 University of Ghana http://ugspace.ug.edu.gh Table 4.14b: Correlation Matrix for Cost Efficiency and Technical Efficiency Variables LOTA AGE LITA BKCN BKZSCOR BOCTA COTI NIM RTBILL GDPG LOTA 1.00 AGE 0.09 1.0 0 LITA (0.80) (0.07) 1.0 0 BKCN 0.15 (0.17) (0.12) 1.0 0 BKZSCOR (0.15) 0.19 0.13 (0.80) 1.0 0 BOCTA 0.15 (0.18) (0.14) 0.69 (0.74) 1.0 0 COTI 0.04 (0.20) (0.08) 0.06 0.00 0.02 1.0 0 NIM (0.22) (0.05) 0.12 (0.15) 0.12 (0.15) (0.21) 1.0 0 RTBILL (0.09) 0.07 0.09 (0.27) 0.43 (0.40) 0.08 0.06 1.0 0 GDPG 0.01 (0.05) 0.01 0.14 (0.16) 0.44 0.10 (0.06) (0.04) 1.0 0 160 University of Ghana http://ugspace.ug.edu.gh Model Diagnostic tests We present empirical results for the determinants of credit union efficiency as specified in equation 3.15. Table 4.9a is the results from the mixed effect model. From Table 4.9a, the Wald chi tests are 301.52, 304.54 and 309.34 with respective probability values of 0.0000 in the case for the cost efficiency models. The Wald chi test in the case for the technical efficiency models are 129.22, 134.22 and 136.66 for the internal model in panel 1, banking sector model and the full model respectively. The probability values are 0.0000 for all the cases as captured in Table 4.9b. The Wald chi test scores for our estimations in Table 4.9a and Table 4.9b support our claim that all the parameters used in the model are jointly significant. In case of the truncated Tobit estimations for cost efficiency and technical efficiency in Table 4.9c, the Likelihood Ratio of 229.79 and 136.66 with probability values of 0.0000 supporting our claim that all the parameters used in the model are jointly significant. 161 University of Ghana http://ugspace.ug.edu.gh Table 4.15a: Cost Efficiency Regression Result (1) (2) (3) VARIABLES Internal model Banking Sector Model Full Model SIZE 0.0633*** 0.0646*** 0.0643*** (9.0200) (8.2205) (8.1846) Z SCORE -0.0002 -0.0003 -0.0003 (-1.1753) (-1.2217) (-1.2203) NWTA 0.0023 0.0024 0.0023 (1.5017) (1.5587) (1.5027) LLP -0.0054*** -0.0054*** -0.0055*** (-2.9932) (-2.9682) (-3.0169) HHI  COMB -0.0856** -0.0855** -0.0803* (-1.9830) (-1.9636) (-1.8309) N IETA -0.0071*** -0.0072*** -0.0071*** (-3.1456) (-3.1248) (-3.1151) L OTA -0.0010 -0.0011 -0.0010 (-1.4977) (-1.5704) (-1.5117) AGE 0.0035*** 0.0036*** 0.0035*** (4.0203) (4.0490) (4.0065) L GLITA 0.1383*** 0.1386*** 0.1383*** (5.9887) (5.9914) (5.9868) BCKN5 -0.0014 -0.0009 (-1.3259) (-0.7782) BKZSCORE -0.0482 -0.0564* (-1.5654) (-1.7462) BKOCTA -0.0068 -0.0085* (-1.6284) (-1.9224) R TBILL -0.0005 (-0.2101) G DP 0.0044 (1.1941) C onstant -0.726 7*** 0.02 05 0.1099 (-4.3675) (0.0423) (0.2241) C redit Unions 61 61 61 Observations 427 427 427 Wald chi2 301.52 304.54 309.34 Prob > chi2 0.0000 0.0000 0.0000 RE Var Residual 0.0263 0.0262 0.0260 (14.6927)*** (15.2326)*** (14.6067)*** t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 There is a direct and significant relations between cost efficiency and size of the credit union. Table 4.9a suggests that increasing the size of the credit union would improve cost 162 University of Ghana http://ugspace.ug.edu.gh efficiency in all the three models presented. The inference here is that, generally, bigger sized or lager credit unions are more cost efficient than smaller sized credit unions as they enjoy economies of scale in their operations. These results on size agree with Esho (2001) who demonstrates that size is a significant determinant of credit union efficiency. The solvency level of the credit union has negative but insignificant relations with cost efficiency. Increasing provision for bad and doubtful debt infers that credit unions are, on the average, most likely not to be efficient, an issue that should be of concern to managers of credit unions. Income diversification activities in non-loan activities has a negative coefficient and significantly associates with cost efficiency, generally increasing the inefficiency of credit unions, a conclusion Worthington (2000) also reaches. Thus, credit unions aiming at cutting down costs must be wary of non-loan income activities especially as it also exposes credit unions to higher levels of risk. Increasing management expense capture by NIETA implies that cost efficiency levels in credit unions would decline relations that are very significant in Table 4.9a. The loan business of the credit union captured by LOTA reveal a negative but insignificant coefficient, implying that high level loans granted from resources can cause economic inefficiencies on the average; in the internal model, a 1 percent increase in loans would result in a 0.0014 decrease in cost efficiency. The increase in Age of the credit union suggests an improvement in cost efficiency as Esho (2001) and Wijesiri, Viganò and Michele (2015) also observe. This may be attributable to experience and learning curve effect as the credit union would not want to repeat decisions that might have led to loss in the past. 163 University of Ghana http://ugspace.ug.edu.gh A longer operating existence gives the credit union the opportunity to improve its cost efficiency. Liquidity implies improved cost efficiency for credit unions in all three models, suggesting that liquid investment, mostly money market investment, has low turn around cost for credit unions. It must also be added that this lower return implies less risk, showing a positive and significant association to cost efficiency. This is a confirmation of the canonical finance investment axiom that high risk goes with high return, and low risk goes with low return. In the banking sector model, all the relationships afore mentioned still hold with the same level of significance. The addition from this model is that increase in solvency of the banking sector would lead to a decline in cost efficiency for credit unions. Solvency of the banking sector which relates to stability may draw a lot of customers and clients to banks. Extra costs would have to be expended to enable the credit unions to stay in business hence the decline in cost efficiency in this regard. As the banking sector’s cost to total asset increases, credit unions stand to become either cost efficient or less so. This can be as a result of the direct competition that exists between credit unions and banks. Additionally, the temptation to win the same customer would mean each competitor would end up spending much more, hence this nature of relationship. Top 5 bank size concentration does little to impact on efficiency. Increasing monopoly by big banks would mean less cost efficient credit unions as costumers would naturally draw to credit unions for business. All banking sector development indicators have negative coefficients to credit union cost efficiencies. From this it is obvious that the banking sector matters in credit union efficiency 164 University of Ghana http://ugspace.ug.edu.gh as also found in Battaglia, Farina, Fiordelisi and Ricci (2010). This can be attributable to the competition that exists between these two competing financial institutions on the financial market. On the economy wide level a negative and insignificant relation exists between cost efficiency of credit unions and real return one year Treasury bill rate. A growing economy can lead to improvement in cost efficiency for credit unions as indicated by the positive coefficient between GDP and cost efficiency. 165 University of Ghana http://ugspace.ug.edu.gh Table 4.15b: Regression Results Technical Efficiency (1) (2) (3) V ARIABLES Internal Model Banking Sector Model Full Model SIZE 0.0185** 0.0185* 0.0180* -1.9769 -1.7653 -1.7215 ZSCORE -0.0002 -0.0002 -0.0002 (-0.7183) (-0.7052) (-0.7050) N WTA -0.0043** -0.0043** -0.0044** (-2.0976) (-2.0709) (-2.1294) LLP -0.0039 -0.0036 -0.0037 (-1.6236) (-1.4714) (-1.5368) HHI COMB 0.2609*** 0.2529*** 0.2614*** -4.526 -4.355 -4.4701 NIETA -0.0008 -0.0013 -0.0012 (-0.2721) (-0.4195) (-0.4008) LOTA -0.0020** -0.0021** -0.0020** (-2.3343) (-2.3123) (-2.2428) A GE 0.0026** 0.0026** 0.0026** -2.1925 -2.2332 -2.1804 LGLITA 0.1668*** 0.1699*** 0.1693*** -5.4077 -5.5066 -5.4959 BKCN5 -0.0008 0.0000 (-0.5389) (-0.0231) B KZSCORE -0.0677* -0.0815* (-1.6479) (-1.8933) B KOCTA -0.0096* -0.0120** (-1.7283) (-2.0461) RTBILL 0.0004 -0.1248 G DP 0.0061 -1.2421 C onstant 0.42 13* 1.430 3** 1.5671** -1.8959 -2.2181 -2.3965 Credit Unions 61 61 61 Observations 427 427 427 Wald chi2 129.22 134.22 136.66 Prob > chi2 0.0000 0.0000 0.0000 RE Var Residual 0.0469 0.0465 0.04626 (14.6151)*** (14.6117)*** (14.6146)*** t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 166 University of Ghana http://ugspace.ug.edu.gh Technical efficiency measures management’s ability to make use of resources and avoid wastage as much as possible. In Table 4.9b, size has significant relations with technical efficiency comparable to Wijesiri, Viganò and Michele (2015) conclusions; it exhibits positively with technical efficiency and bigger size credit unions are more technically efficient in all three models. The higher the new worth to asset, the less technically efficient the credit union can be on the average. The relationship between HHI  that is income diversification from non-loan activities COMB and technical efficiency is positive as an increase in credit unions’ investment in non-loan income leads to an increase in technical efficiency. It must be explained that these non-loan investments need a lot of management commitment; less technically efficient credit unions may stick to loan business and make less ventures into non-loan activities because they may not have the capacity to meander their way in managing these investments. Management quality therefore matters. The more credit unions spend on non-interest expense, NIETA, the less technically efficient that credit union is on the average; if not, they should spend less on management expense. Loan to asset of credit unions shows an inverse relation with technical efficiency. Inherent is the credit business is bad loans, increasing loans comes with commensurate bad loans, loan increase implies a 0.0020 decrease on the average in technical efficiency in all case. A similar relation also hold for bad loans, increasing bad loans leads to higher levels of inefficiencies on the part of management of the credit union. The longevity of the credit union’s operation also matters; an increase in years of existence of credit union would imply 167 University of Ghana http://ugspace.ug.edu.gh more technical efficiency, the reason being that these older credit unions may have learned from experience. The more stable the banking sector, as measure by the Zscore, the less technically efficient the credit unions become, probably because labor may then move to banks, denying the credit union of quality management. An increase in the banking sector’s efficiency through an increase in banking cost to income ratio implies a decline in technical efficiency of credit unions. Top 5 bank concentration has a negative relation to technical efficiency in the banking sector model and positively related to technical efficiency in the full model; however both are not significant in the relations. An inefficient banking sector hardly pushes credit unions to pursue improvement in technically efficiency. Real Treasury bill and GDP growth are all positively related to credit union technical efficiency yet not significantly. 168 University of Ghana http://ugspace.ug.edu.gh Table 4.15c: Tobit Regression Results (1) (2) VARIABLES Cost Efficiency Technical Efficiency SIZE 0.0665*** 0.0252** (8.0939) (2.1169) ZSCOR E -0.0003 -0.0003 (-1.3359) (-0.8953) NWTA 0.0024 -0.0053** (1.505) (-2.2361) LLP -0.0057*** -0.0042 (-3.0091) (-1.5532) HHI COMB -0.0868* 0.2866*** (-1.8984) (4.3591) NIETA -0.0071*** -0.0011 (-2.9708) (-0.3032) LOTA -0.0011 -0.0021** (-1.5105) (-2.1338) AGE 0.0038*** 0.0028** (4.116) (2.0964) LGLIT A 0.1411*** 0.1829*** (5.8632) (5.3171) BKCN 5 -0.001 -0.0003 (-0.8089) (-0.1813) BKZSC ORE -0.0592* -0.0999** (-1.7620) (-2.0706) BKOC TA -0.0090* -0.0143** (-1.9491) (-2.1700) RTBIL L -0.0005 0.0005 (-0.2172) (0.1359) GDP 0.0044 0.0072 (1.147) (1.3197) Consta nt 0.1235 1.7980** -0.2419 -2.4436 LR chi 2(14) 229.79 136.66 Prob > chi2 0.0000 0.0000 Credit Unions 61 61 Observations 427 427 t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 169 University of Ghana http://ugspace.ug.edu.gh A two limit Tobit regression results is presented in Table 4.9c. The range cost efficiency scores are scaled from 0.1078 to 1.000 and for technical efficiency scaled from 0.1102 to 1.000 for both lower limit and upper limit respectively. In these results we see that if credit unions increase their size by 1 percent the expected size effect would increase 0.0665 percent and 0.0252 percent on cost and technical efficiency respectively while holding all other variables in the model constant. For every 1 percent increase in NWTA, there exist 0.0024 increase in cost efficiency estimation and 0.0053 decrease in NWTA in the technical efficiency estimation. Increase loan loss would lead to a decrease in the expected value by 0.0057 in cost efficiency. The expected values of non-loan income diversification activities decrease by 0.0868 in the cost efficiency model, and a 0.2866 increase in the technical efficiency for every 1 percent increase in HHI  these relations are significant. Non-interest expense to total asset COMB ; exhibit a decrease of 0.0071, for every 1 percent increase in cost efficiency. For every increase in NIETA the relationship is insignificant in technical efficiency with a coefficient of -0.0011. The loan business exhibits an insignificant relation and a reduction in the cost efficiency of 0.0011l but a significant relation of -0.0021 for the technical efficiency. An increase in Age by 1 year implies an increase in value as given in the coefficient under both cost efficiency and technical efficiency. Liquidity captured by LITA shows an increase in expected value of 0.1411 under cost efficiency and 0.1829 under technical efficiency, on the average, from a 1 percent increase in LITA. 170 University of Ghana http://ugspace.ug.edu.gh The stability of the banking sector shows a statistically significant decline of 0.0592 and 0.0999 decline in expected values for a unit increase in the case of the two dependent variables respectively in table 4.9c. The efficiency level of the banking sector imply negative relations with cost efficiency and technical efficiency for a 1 percent increase in banking sector overhead cost leads to an decrease of 0.0090 and 0.0143 in expected value of cost efficiency and technical efficiency respectively. The real Treasury bill rate and GDP growth do not exhibit statistically significant relations with cost and technical efficiency as presented in Table 4.9c. 4.6. Integrated performance of the Credit Union Objective 8; empirically assess the determinants credit union financial performance in an integrated model. In Table 4.10a and Table 4.10b we present results from the estimation of equation 3.16. In this model, we assess the financial performance of the credit union in which our diversification measures, lending and the estimated cost efficiency and technical efficiency estimates is employed in the integrated model. This approach is different from Goddard, McKillop and Wilson’s 2008 assessment of credit unions performance in the United States of America. The difference between Table 4.10a and 4.10b is that cost efficiency is introduced in place of cost to income, whiles cost to income is maintained in the technical efficiency model in Table 4.10b as we believe the input prices used in estimating cost 171 University of Ghana http://ugspace.ug.edu.gh efficiency does well in capturing cost of operations in the credit union but not in the case of technical efficiency in the results in Table 4.10b. Model Diagnostic tests We present empirical results for the integrated credit union financial performance as specified in equation 3.16. Both Table 4.10a and Table 4.10b is from the mixed effect model. From Table 4.10a, the Wald chi tests are 938.30 961.44 920.58 936.91 and 959.14 with respective probability values of 0.0000 in the case for the cost efficiency models. The Wald chi test in the case for the technical efficiency models are 1756.14, 1753.45, 1727.78, 1730.19, 1728.98 and 1729.34. The probability values are 0.0000 for all the cases as captured in Table 4.10b. The Wald chi test scores for our estimations in Table 4.10a and Table 4.10b support our claim that all the parameters used in the model are jointly significant. The same model is presented in Appendix 4a and 4b, where the regression estimate is without a constant term; generally the conclusions and the inference are not much different for the two estimation techniques adopted. From Table 4.11a, SIZE is significant under model 2 of the no constant model. Also NWTA is statistically significant under all models estimated in the no constant term equation a situation not so under the estimation with a constant term. Finally LLP is not significant in all models under the no constant model, however in the estimates with a constant term is only in case of HHI  that BKCN3 is LFI insignificant. The conclusion drawn for the results in Table 4.10b is pretty much the same in 172 University of Ghana http://ugspace.ug.edu.gh no constant results presented in Table 4.11b, only for the case of NWTA, which was not significant under model with a constant term, but is significant under the no constant term. Technical efficiency is not significant in model 5 of Table 4.10b, but in the case of the no constant model, technical efficiency is not significant in model 3, 4 and 5. Table 4.10a shows that increasing diversification income with the non interest category of income for credit unions is significant at 5 percent and negatively associates with risk adjusted return on equity and return on asset with coefficients values of 0.7337 and 0.7115. From the relation between non-financial income category of income we can infer that it would be best suited for income smoothing purpose within the income diversification argument. Increased credit union loan business, that is Loan to asset, would positively influence on financial performance for credit unions and this relation holds also in the case of increasing diversification income in liquid-financial investment and the increase combined diversification income as presented in Table 4.10a. Increasing cost efficiency in credit union’s production imply more profitability as evidenced in the positively and highly statistically significant coefficient of 1.6829, 1.7205, 1.3302, 1.3678, 1.4531 and 1.4949 respectively. Top 3 asset bank concentration also indicates a positive significant relations with risk adjusted return on equity and return on asset in the presence of income diversification within the non interest category and the combined non- loan income diversification category. No statistical case can be made for the case of diversification within the liquid-financial investment group. 173 University of Ghana http://ugspace.ug.edu.gh Table 4.16a: Integrated Cost Efficiency Financial Performance (1) (2) (3) (4) (5) (6) VARIABLES RAROE RAROA RAROE RAROA RAROE RAROA SIZE 0.4297*** 0.4255*** 0.5506*** 0.5444*** 0.5378*** 0.5306*** (2.9373) (2.9032) (3.9480) (3.8994) (3.8863) (3.8256) LOTA 0.0166** 0.0166** 0.0197*** 0.0196*** 0.0188*** 0.0188*** (2.3793) (2.3819) (2.8704) (2.8484) (2.7519) (2.7478) N IM 0.0500*** 0.0504*** 0.0494*** 0.0501*** 0.0525*** 0.0527*** (8.2273) (8.2757) (7.9659) (8.0580) (8.3778) (8.3930) L ITA 0.0206*** 0.0214*** 0.0285*** 0.0289*** 0.0256*** 0.0264*** (2.8129) (2.9215) (4.3592) (4.4180) (3.8776) (3.9870) LLP -0.0719*** -0.0705*** -0.0745*** -0.0733*** -0.0712*** -0.0699*** (-6.3381) (-6.2059) (-6.5030) (-6.3918) (-6.2597) (-6.1344) N WTA -0.0191* -0.0097 -0.0206* -0.0112 -0.0189* -0.0096 (-1.6760) (-0.8520) (-1.8024) (-0.9780) (-1.6606) (-0.8419) Z SCORE 3.6700*** 3.7302*** 3.6708*** 3.7304*** 3.6393*** 3.7013*** (23.9148) (24.2617) (23.7607) (24.1186) (23.6039) (23.9541) BKCN3 0.0397* 0.0390* 0.0377 0.0375 0.0394* 0.0386* (1.7326) (1.6977) (1.6297) (1.6192) (1.7185) (1.6798) I NFL -0.0128 -0.0108 -0.0109 -0.0094 -0.0106 -0.0087 (-0.5601) (-0.4725) (-0.4745) (-0.4066) (-0.4673) (-0.3815) GDP -0.0319 -0.0292 -0.0378 -0.0354 -0.0306 -0.0282 (-1.1603) (-1.0628) (-1.3669) (-1.2780) (-1.1133) (-1.0235) CE 1.6829*** 1.7205*** 1.3302*** 1.3678*** 1.4531*** 1.4949*** (4.3081) (4.3962) (3.5603) (3.6567) (3.9308) (4.0351) HHI  NFI -0.7337** -0.7115** (-2.4325) (-2.3546) HHI LFI  0.2043 0.2722 (0.5443) (0.7244) HHI COMB 0.7294** 0.6863** (2.3399) (2.1968) Constant -4.2671*** -5.2449*** -5.2307*** -6.2219*** -5.7236*** -6.6397*** (-2.5846) (-3.1709) (-3.1918) (-3.7923) (-3.4992) (-4.0505) Observations 427 427 427 427 427 427 Credit Unions 61 61 61 61 61 61 W ald Chi 2 938.30 961.44 920.58 945.33 936.91 959.14 Prob> chi2 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 In table 4.10b we find that increase within diversification income from interest income associates negatively with coefficient 0.5058 and 0.5392 risk adjusted return on equity and risk adjusted return on asset. Improvement in technical efficiency of credit unions would 174 University of Ghana http://ugspace.ug.edu.gh lead to an increase of 0.4396 and 0.4501 in risk adjusted return on equity and risk adjusted return on equity. Liquid financial income positively but not significantly with technical efficiency but is significant and positive, under combined non loan income diversification, technical efficiency is positive and significant under risk adjusted return on asset and not equity, whiles loan to asset is still positive on credit union performance. Macroeconomic indicators in the form of price levels and worth in the economy negatively associates with credit union performance albeit not significant. 175 University of Ghana http://ugspace.ug.edu.gh Table 4.16b: Integrated Technical Efficiency Financial Performance (1) (2) (3) (4) (5) (6) VARIABLES RAROE RAROA RAROE RAROA RAROE RAROA SIZE 0.3830*** 0.3906*** 0.4246*** 0.4301*** 0.4297*** 0.4329*** (3.9829) (4.0178) (4.4576) (4.4711) (4.4900) (4.4782) COTI -0.0277*** -0.0274*** -0.0275*** -0.0272*** -0.0274*** -0.0271*** (-16.6100) (-16.2590) (-16.3794) (-16.0490) (-16.3012) (-15.9891) LOTA 0.0110** 0.0112** 0.0132** 0.0132** 0.0132** 0.0132** (1.9718) (1.9858) (2.3838) (2.3604) (2.3885) (2.3784) N IM 0.0235*** 0.0240*** 0.0237*** 0.0245*** 0.0241*** 0.0244*** (4.7651) (4.8184) (4.6842) (4.7918) (4.7222) (4.7340) L ITA 0.0134** 0.0146** 0.0186*** 0.0193*** 0.0185*** 0.0194*** (2.2802) (2.4558) (3.3685) (3.4659) (3.3565) (3.4934) LLP -0.0063 -0.0060 -0.0067 -0.0067 -0.0064 -0.0061 (-0.6462) (-0.6107) (-0.6781) (-0.6684) (-0.6482) (-0.6132) NWTA -0.0124 -0.0030 -0.0143 -0.0048 -0.0140 -0.0045 (-1.3553) (-0.3226) (-1.5529) (-0.5179) (-1.5221) (-0.4866) Z SCORE 3.5220*** 3.5823*** 3.5308*** 3.5904*** 3.5241*** 3.5861*** (29.0336) (29.2117) (28.9298) (29.1344) (28.7521) (28.9618) BKCN 0.0239 0.0233 0.0218 0.0219 0.0218 0.0213 (1.3142) (1.2709) (1.1900) (1.1828) (1.1961) (1.1531) INFL -0.0051 -0.0032 -0.0038 -0.0023 -0.0033 -0.0015 (-0.2849) (-0.1741) (-0.2069) (-0.1232) (-0.1830) (-0.0801) G DP -0.0083 -0.0058 -0.0135 -0.0112 -0.0117 -0.0093 (-0.3818) (-0.2649) (-0.6190) (-0.5069) (-0.5332) (-0.4217) HHI NFI  -0.5392** -0.5058** (-2.3926) (-2.2201) HHI  LFI 0.0953 0.1701 (0.3205) (0.5669) HHI COMB 0.1466 0.0994 (0.5841) (0.3921)  TE 0.4396* 0.4501** 0.3859* 0.3906* 0.3746 0.3948* (1.9358) (1.9603) (1.6755) (1.6793) (1.6219) (1.6921) Constant -0.9420 -2.0356 -1.3311 -2.4342* -1.4522 -2.4725* (-0.7381) (-1.5779) (-1.0383) (-1.8804) (-1.1128) (-1.8755) Observations 427 427 427 427 427 427 Credit Unions 61 61 61 61 61 61 Wald Chi 2 1756.14 1753.45 1727.78 1730.19 1728.98 1729.34 Prob> chi2 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 t-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 Bad debt plays a significant role in the presence of cost efficiency but not in technical efficiency. So is NWTA in the return on equity model with the cost efficiency measure used as an explanatory variable. Cost efficiency is important in all types of non-loan income and 176 University of Ghana http://ugspace.ug.edu.gh overall profitability for credit unions. The evidence in this study points to a positive relation. By improving cost efficiency, credit unions stand to record higher performance on a risk adjusted basis. 4.7. Hypothesis Testing In analysing the results generated, the study accepts the null hypothesis that increasing liquid diversification income positively associates with risk adjusted performance of credit unions. A similar conclusion is made for non-financial income diversification, where the null hypothesis that non-financial income relates negatively with financial performance is accepted. Financial Performance Cost Efficiency (+) Technical Efficiency (+) (-) (-) (-) (+) (-) Loan Income (+) Non- Loan Income (+) Liquid Financial Investment Income (+) Non -Interest Income (-) Figure 5.1 Established relationship in the credit union performance Nexus 177 University of Ghana http://ugspace.ug.edu.gh Similarly the null hypothesis that non-loan income positively related to risk adjusted performance in the credit union is also accepted, as is the null hypothesis that non loan income negatives associates cost efficiency. However, the study rejects the null that non- loan income negatively relates to technical efficiency. Instead, it accepts the null in the last hypothesis that, non-loan income, loan to asset and efficiency significantly impact credit union risk adjusted performance. These relationship has been displayed in figure 5 based on the thematic areas analysed from the results estimated from the empirical models used in this thesis. 178 University of Ghana http://ugspace.ug.edu.gh CHAPTER FIVE SUMMARY OF FINDINGS, CONCLUSION AND RECOMMENDATION 5.1 Introduction This thesis analyzed the level of income diversification and how income diversification associates with credit union financial performance. It also analyzed how discretional and non-discretional factors relate to lending in credit union. We also estimated cost and technical efficiency of credit union and its determinants in Ghana. Finally the thesis incorporates all these objectives into a financial performance model. This final chapter offers a summary of the findings, the contributions of the study, conclusions, policy recommendations as well as limitations and suggestions for further research emanating from this thesis. 5.2 Summary of Findings  Per the study’s objective for income diversification, we have established that non- loan income comprised a good level of the income basket for credit unions. There exist a fair spread of diversification within the non-financial income category of the credit union, compared the higher level of concentration of income within the liquid- financial investment category.  This study also posits that non-financial income is best suited for income smoothening within the credit union total income loan income-non-loan income 179 University of Ghana http://ugspace.ug.edu.gh diversification argument. Liquid-financial investment income positively associates with risk adjusted performance. The overall non-loan income diversification relates positively with financial performance of the credit union.  The study shows that discretional factor tend to influence credit unions’ loan business greatly. The size of the credit union greatly influences credit unions’ lending business, positively to a point and then negatively after an optimum level. Increasing the lending rates in the credit union would be inimical to the loan portfolio of the credit union. Increasing spread within non-loan diversification reduces credit union lending. The development of the banking sector, especially in regard to the dominance of the banking industry by the top 3 banks, positively influences credit unions’ loan business.  The study further reveals that a concentrated banking sector tends to increase the demand for loans in the credit union. Increasing the rate on treasury bills as a way of reducing liquidity in the economy by monetary authority does not help in credit union lending. In general growth in the economy does not further help credit unions’ business.  The efficiency levels in credit unions point to the fact that credit unions are more technically efficient than cost efficient on the average. Overall, credit unions have improved their cost efficiency over the 7 year period compared to improvement levels in technical efficiency. We observe that credit union technical efficiency 180 University of Ghana http://ugspace.ug.edu.gh hardly translates into cost efficiency. In the case of cost efficiency increasing spread in combined non loan income reduces the production efficiency of credit unions. On the other hand, the increasing spread in combined non-loan income increases the technical efficiency of credit unions. Finally, making more loans in the credit union indicates a decline in both cost and technical efficiency.  From the banking industry, we have also established that banking stability does not positively connect with cost and technical efficiency in credit unions. Increasing inefficiency of the banking sector also does not motivate cost and technical efficiency in the credit union. We can assign the competition reason for this nature of relationship existing between the banking industry and the credit union industry that as banks become inefficient, the demand for financial services from the credit union increase. This has the possibility of hampering efficiency in the credit union.  Finally from our integrated performance assessment, this study posits that cost efficiency positively associates with risk adjusted returns on equity and risk adjusted return on asset in all our estimations. Non-financial income negatively associates with the financial performance of credit unions, while income of liquid-financial investment positively links with financial performance just as for combined non loan income. Increasing loan to asset of the credit union improve risk adjusted performance. The same nature of relationship holds in the integrated model on financial performance when technical efficiency is used in place of cost efficiency. Technical efficiency is significant and positive in all our performance estimation 181 University of Ghana http://ugspace.ug.edu.gh with the execution of risk adjusted return on equity with combined non loan diversification income capture in the model although technical efficiency maintains its positive relations. 5.3 Contributions of the Study Firstly we have added to existing knowledge on the income diversification debate by drifting from the know area of how diversification affects financial performance both on a risk adjusted and non-risk adjusted performance by focusing on the how credit union specific factors associates in income types with the non-loan income group which is an attempt to push our understanding of income diversification since, non-loan income would always play a vital role in the income basket of credit unions in particular and other financial institution in general. In doing this we add to the existing evidence that size, efficiency, net worth, loan to asset, banking sector concentration, inflation, growth of the economy and workplace and community ownership type of credit union significantly associates with credit union non loan income Secondly we have contributed to existing literature that there exist an inverted U-shaped relation between size of the credit union and increase in loan portfolio. Increasing loan rate is not a good option for credit union desiring increase in loans. A concentrated banking sector creates the needed opportunity for credit union to increase loan portfolio. The pursuit of non-loan income hampers credit union loan business. Restrictive or contractionary monetary policy using Treasury bill open market operation restricts the lending business of 182 University of Ghana http://ugspace.ug.edu.gh credit union. GDP growth does not translate into a positive loan making business for the credit union. The study has succeeded in adding to the position that technical efficiency does not necessarily translate into cost efficiency in the credit union setting and that credit union are more technically efficient than cost efficient. Non-loan income does not improve cost efficiency in the credit union. The banking sector of the economy does not improve efficiency in the credit union. Longevity positively associates with efficiency in the credit union. The pursuit of non-financial implies improvement in technical efficiency. The macroeconomic environment does not significantly associate with credit union efficiency. The study further contributes that non-loan income impacts risk adjusted performance of the credit union negatively in reference to non-financial income and positive relation with combined non-loan income. Improving cost efficiency translates into improved performance. Increasing loan portfolio means improved financial performance. Increased concentration within the liquid-financial investment positively associates with credit union albeit not significantly. 5.4 Conclusion This study has established that although, non-loan income is not the core factor in the setup of credit unions, it currently plays, and will continue to play an important role in the income 183 University of Ghana http://ugspace.ug.edu.gh basket of credit unions. The study finds that within the non-loan income category, non- financial income is better positioned to be used as a strict diversification strategy, a conclusion also established in the financial institution diversification literature. Funds allocation for liquid-financial investment is most likely to be for supplementary income purpose and a quick fix for credit unions’ unanticipated loan demand as they can be easily liquidated at a relatively low cost. Smaller sized credit unions stand to benefit more from income diversification. Increased investment in liquid asset and resource usage hampers how much credit unions can generate from non-financial income, whiles ownership type play the least role in non- financial income. Credit unions’ operational longevity and their stock of liquid investment facilitate the diversification of liquid-financial investment. On the other hand, net interest margin tends to restrict how far credit unions can diversify their liquid-financial investment. For the combined non-loan income, the credit union’s age is critical, net interest margin is important, while size, liquid investment and expense do have some managerial implications for the credit union’s income diversification drive. Loan creations would continue to play a significant role in credit unions’ existence, since one of the main reasons for establishing credit unions is to offer loans at comparable and competitive rates on the financial market. Attempts by credit union managers to diversify income sources would mean a reduction in the credit unions’ loan granting ability. Loan loss management is also a prime issue to be considered by the managers of the credit union 184 University of Ghana http://ugspace.ug.edu.gh although the consignor effect can be relied on to manage loan asset which goes a long way to improve credit unions’ loan extension activities. Managers of credit unions would have to be wary of increasing lending rates as it can be detrimental to the wellbeing of the credit union since it is likely to drive loan business away from the union to other competing financial institutions. Expense management and efficiency would have to be the main focus so the managers of credit unions could pass on inefficiency onto the borrower by charging higher rates on loans or paying lower deposit rates or lower dividends on shares, a condition which would in the long run take away business form the credit union and destroy the very reason for its establishment. The current study estimates cost and technical efficiency for credit unions and investigates efficiency for Ghanaian credit unions during the period 2008 to 2014. Credit unions operated at cost efficiency average of 38.9percent, while technical efficiency averaged 54.4percent for the period. A greater number of credit unions in the study are not cost efficient; the opposite, meanwhile, holds true for technical efficiency. The results show that many credit unions are on the average technically efficient. We also realized that cost efficiency does not necessary translate into technical efficiency as there exits varying scores and positions in both measures of efficiency score for the same credit union among the sample credit unions. 185 University of Ghana http://ugspace.ug.edu.gh From the mixed effect and the two limit Tobit regression estimates, efficiency of the credit union is more internally driven by factors like size, net worth to asset, bad loans, non-loan income, non-interest expense, loan to asset liquidity and Age. Credit union efficiency is also significantly associated with competition and efficiency of the banking sector of the economy. Banking sector development displayed in this study mostly exhibited negative relation with credit union efficiency. The wider economy has managerial implication for credit unions’ efficiency but did not bear statistically significant relations. 5.5a. Recommendation Credit unions should continue to concentrate on their core mandate which is to do business with their member owners by providing loans to member borrowers, and aim at reducing operational expense which would shore up income and culminate in high deposit rates for saver members. For the management of credit unions, diversifying income sources is important but should be prioritized in the value creation option only in the case of excess supply of deposit juxtaposed with good control over expenses and low demand for loan facilities. In sum, loan business of the credit union is more of an internal issue, with non-discretional factors having strong influence. In pursing loan business, the credit union management quality matters and should be upheld if loan loss is to be reduced. The environmental issues that should be of concern to managers include the banking activities of big banks and the over cost of banking institutions. An increasing overhead cost of banks would mean the credit unions would have to be prepared to meet increase in loan demand; increasing loan 186 University of Ghana http://ugspace.ug.edu.gh portfolio should be done tactfully by managers of the credit union so as not to experience any diseconomies of scale from the loan business. From the evidence provided in this study, managers of credit unions must improve their production aspects of operations to cut down production cost. Managers must also aim at deploying every resource fully to avoid waste in their efforts to create value for the owners. Understanding banking sector developments, especially that large banks’ increasing dominance of the banking sector and increases in bank overhead costs can negatively affect credit unions efficiency, is paramount. It is therefore appropriate for managers to monitor these aspects of the banking industry and craft strategies that would help improve their efficiency which would then translate into higher levels of returns for owners of these unions. 5.5 b. Policy Recommendation Lastly, a monopolized banking sector and inefficient banking sector does not challenge efficiency improvement in the credit unions industry. It is recommended that policy makers for the financial institutions industry in this instance the monetary policy committee of the central bank look beyond the banking sector and consider the effect of their actions on smaller deposit taking financial institution like the credit union, since from this study we see that a contractionary open market operations affect credit union leading. 187 University of Ghana http://ugspace.ug.edu.gh In regards to efficiency, the central bank would have to employ policies that would reduce the market power of big banks, as this study has revealed that a concentrated banking sector does not spur efficiency in the credit unions. Policies that would make the banking sector more competitive should be pursued by the central bank, as these policies would indirectly improve cost efficiency and technical efficiency in credit unions. 5.6 Limitations and Suggestions for Further Research This thesis has attempted to provide some useful contribution in the evaluation of credit union financial performance, lending business and efficiency, and proffered an integrated approach to assessing credit union performance. Some limitations do exist in our study, including the inability to collect data on a critical mass of credit unions operating in Ghana and the focus on Ghanaian credit union restricts the evidence to Ghana. In this regard an expanded data set study on credit unions in Ghana is suggested. Indeed, if possible, research should be conducted on a sub-regional and also continental level to further enrich our understanding of the credit union. Further research could consider revenue efficiency and how competition from the banking sector of the economy imparts on revenue mobilization of the credit union as we have established that banking sector development influence cost and technical efficiency of credit unions in Ghana. The competitor base can be broaden to accommodate competition also from the other thrifts institutions operating in the economy. 188 University of Ghana http://ugspace.ug.edu.gh Last, but not least, other studies can investigate how technical efficiency can translate into cost efficiency for the credit union sector of the economy. Additionally, variation in credit union ownership type and how that could influence efficiency might be considered in future research. 189 University of Ghana http://ugspace.ug.edu.gh REFERENCES Acharya, V. V., Hasan, I., & Saunders, A. (2006). Should banks be diversified? Evidence from individual bank loan portfolios. The Journal of Business, 79(3), 1355-1412. Acharya, V., I. Hasan, & A. Saunders, (2002). ‘Should banks be diversified? Evidence from individual bank loan portfolios’, Bank for International Settlements, Working Paper, No.118. Adusei, M., & Appiah, S. (2011). Determinants of group lending in the credit union industry in Ghana. Journal of African Business, 12(2), 238-251. Adusei, M. (2013). Determinants of credit union savings in Ghana. Journal of International Development, 25(1), 22-30. Adusei, M. (2015). Bank profitability: Insights from the rural banking industry in Ghana. Cogent Economics & Finance, 3(1), 1078270. Alkhathlan, K. A., & Malik, S. A. (2010). Are Saudi Banks Efficient? Evidence using Data Envelopment Analysis (DEA). International Journal of Economics and Finance, 2(2), 53. Amidu, M. (2014), What Influences Banks Lending in Sub-Saharan Africa? Journal of Emerging Market Finance 13 (1) 1–42. Aggarwal, R.K., Samwick, A.A., 2003. Why do managers diversify their firms? Agency reconsidered. Journal of Finance 58 (1), 71–118. Baele, L., De Jonghe, O., & Vander Vennet, R. (2007). Does the stock market value bank diversification? Journal of Banking & Finance, 31(7), 1999-2023. 190 University of Ghana http://ugspace.ug.edu.gh Banker, R.D., Charnes, A., & Cooper, W.W. (1984). Some models for the estimation of technical and scale inefficiencies in Data Envelopment Analysis. Management Science, 30, 1078–1092. Battaglia, F., Farina, V., Fiordelisi, F., & Ricci, O. (2010). The efficiency of cooperative banks: the impact of environmental economic conditions. Applied Financial Economics, 20(17), 1363-1376. Batten, J. A., & Vo, X. V. (2016). Bank risk shifting and diversification in an emerging market. Risk Management, 18(4), 217-235. Bauer, K. (2008). Detecting abnormal credit union performance. Journal of Banking & Finance, 32(4), 573-586. Berger, A. N., & Humphrey, D. B. (1991). The dominance of inefficiencies over scale and product mix economies in banking. Journal of Monetary Economics, 28(1), 117-148. Berger, A. N., & Udell, G. F. (2002). Small business credit availability and relationship lending: The importance of bank organisational structure. The Economic Journal, 112 (477), F32-F53. Berger, A. N., Demsetz, R. S., & Strahan, P. E. (1999). The consolidation of the financial services industry: Causes, consequences, and implications for the future. Journal of Banking & Finance, 23(2), 135-194. Berger, A. N., Hasan, I., & Zhou, M. (2010). The effects of focus versus diversification on bank performance: Evidence from Chinese banks. Journal of Banking & Finance, 34(7), 1417-1435. 191 University of Ghana http://ugspace.ug.edu.gh Berger, A. N., & Humphrey, D. B., (1997). Efficiency of Financial Institutions: International Survey and Directions for Future Research. European Journal of Operational Research, 98(2), 175–212. Berger, A. N., Miller, N. H., Petersen, M. A., Rajan, R. G., & Stein, J. C. (2005). Does function follow organizational form? Evidence from the lending practices of large and small banks. Journal of Financial Economics, 76(2), 237-269. Berger, A. N. & Udell, G. F., (2006). A more complete conceptual framework for SMEs finance. Journal of Banking and Finance 30, 2945-2966. Berger, A.N. (1993). Distribution-free estimates of efficiency in the U.S. banking industry and tests of the standard distributional assumptions. Journal of Productivity Analysis. 4(3), 61–92. Berger, P. G., & Ofek, E. (1996). Bust up takeovers of value‐destroying diversified firms. The Journal of Finance, 51(4), 1175-1200. Bernanke, B.S. & Gertler, M., (1995). Inside the black box: The credit channel of monetary policy transmission. Journal of Economic Perspectives 9, 27–48. Bernanke, B.S, & Blinder, A., (1988). Credit, money and aggregate demand. American Economic Review, Papers and Proceedings. 78(1). 435-439. Berthoud, R. & Hinton, T., (1989), Credit Unions in the United Kingdom, Policy Studies Institute, Printer Publishers Limited (UK). Bliss, R. T., & Rosen, R. J. (2001). CEO compensation and bank mergers. Journal of Financial Economics, 61(1), 107-138. Bolton, P., & Scharfstein, D. S. (1996). Optimal debt structure and the number of creditors. Journal of Political Economy, 104(1), 1-25. 192 University of Ghana http://ugspace.ug.edu.gh Bolton, P., Freixas, X., Gambacorta, L., & Mistrulli, P. E. (2016). Relationship and transaction lending in a crisis. Review of Financial Studies, hhw041. Boot, A. W., & Thakor, A. V. (1994). Moral hazard and secured lending in an infinitely repeated credit market game. International Economic Review, 899-920. Booz, A. Hamilton. (1985). Diversification: A Survey of European Chief Executives. Booz, Allen and Hamilton. Inc., NY. Brighi, P., & Venturelli, V. (2014). How do income diversification, firm size and capital ratio affect performance? Evidence for bank holding companies. Applied Financial Economics, 24(21), 1375-1392. Burns, K. J. (1998). Focus on Quantitative Methods-Beyond Classical Reliability: Using Generalizability Theory to Assess Dependability. Research in Nursing and Health, 21, 83-90. Busch, R. and Kick., T. (2009), Income Diversification in the German Banking Indistry”, Bundesbank Discussion Paper, No. 09/2009. Camanho, A. S., & Dyson, R. G. (2005). Cost efficiency, production and value-added models in the analysis of bank branch performance. Journal of the Operational Research Society, 56(5), 483-494. Casu, B., Dontis‐Charitos, P., Staikouras, S., & Williams, J. (2016). Diversification, size and risk: The case of bank acquisitions of nonbank financial firms. European Financial Management, 22(2), 235-275. Cecchetti, S.G. (1999), ‘Legal Structure, Financial Structure, and the Monetary Policy Transmission Mechanism’, Economic Policy Review, 5(2): 9–28. 193 University of Ghana http://ugspace.ug.edu.gh Charnes, A., Cooper, W. W & Rhodes, E., (1978). Measuring the Efficiency of Decision Making Units. European Journal of Operational Research, 2(6), 429–444. Chiorazzo, V., Milani, C., & Salvini, F. (2008). Income diversification and bank performance: Evidence from Italian banks. Journal of Financial Services Research, 33(3), 181-203. Coelli, T. (1996). A Guide to DEAP Version 2.1: A Data Envelopment Analysis (Computer Program).Working Paper. CEPA,University of New England, Armidale. CUA , Credit Union Association Ghana (1998). News Letter, Accra, Ghana. Damankah, B. S., Anku-Tsede, O., & Amankwaa, A. (2014). Analysis of Non-Interest Income of Commercial Banks in Ghana. International Journal of Academic Research in Accounting, Finance and Management Sciences, 4(4), 263-271. De Jonghe, O. (2010). Back to the basics in banking? A micro-analysis of banking system stability. Journal of Financial Intermediation, 19(3), 387-417. De Vries, C. G. (2005). The simple economics of bank fragility. Journal of Banking & Finance, 29(4), 803-825. Debreu, G. (1951). The coefficient of resource utilization. Econometrics, 19(3):273–292. Demirgüç-Kunt, A., & Huizinga, H. (2010). Bank activity and funding strategies: The impact on risk and returns. Journal of Financial Economics, 98(3), 626-650. Demsetz, R. S., & Strahan, P. E. (1997). Diversification, size, and risk at bank holding companies. Journal of Money,Credit, and Banking, 300-313. Deng, S. E., & Elyasiani, E. (2008). Geographic diversification, bank holding company value, and risk. Journal of Money, Credit and Banking, 40(6), 1217-1238. 194 University of Ghana http://ugspace.ug.edu.gh Denis, D. J., Denis, D. K., & Sarin, A. (1997). Agency problems, equity ownership, and corporate diversification. The Journal of Finance, 52(1), 135-160. DeYoung, R., & Rice, T. (2004). Noninterest income and financial performance at US commercial banks. Financial Review, 39(1), 101-127. DeYoung, R., & Roland, K. P. (2001). Product mix and earnings volatility at commercial banks: Evidence from a degree of total leverage model. Journal of Financial Intermediation, 10(1), 54-84. DeYoung, R., & Torna, G. (2013). Nontraditional banking activities and bank failures during the financial crisis. Journal of Financial Intermediation, 22(3), 397-421. Dong, Y., Hamilton, R., & Tippett, M. (2014). Cost efficiency of the Chinese banking sector: a comparison of stochastic frontier analysis and data envelopment analysis. Economic Modelling, 36, 298-308. Doumpos, M., Gaganis, C., & Pasiouras, F. (2016). Bank diversification and overall financial strength: International evidence. Financial Markets, Institutions & Instruments, 25(3), 169-213. Edirisuriya, P., Gunasekarage, A., & Dempsey, M. (2015). Australian specific bank features and the impact of income diversification on bank performance and risk. Australian Economic Papers, 54(2), 63-87. Ehrmann, M., Gambacorta, L., Martinez‐Pagés, J., Sevestre, P., & Worms, A. (2003). The effects of monetary policy in the euro area. Oxford Review of Economic Policy, 19(1), 58-72. Ely, D. P., & Robinson, K. J. (2009). Credit unions and small business lending. Journal of Financial Services Research, 35(1), 53. 195 University of Ghana http://ugspace.ug.edu.gh Emmons, W. R., & Schmid, F. A. (2000). Bank competition and concentration: Do credit unions matter?. Federal Reserve Bank of St. Louis Review, (May), 29-42. Esho, N. (2001). The determinants of cost efficiency in cooperative financial institutions: Australian evidence. Journal of Banking & Finance, 25(5), 941-964. Esho, N., Kofman, P., & Sharpe, I. G. (2005). Diversification, fee income, and credit union risk. Journal of Financial Services Research, 27(3), 259-281. Fairbairn, B. (1994). The meaning of Rochdale: The Rochdale pioneers and the co-operative Principles. Faleye, O., & Krishnan, K. (2017). Risky lending: Does bank corporate governance matter?. Journal of Banking & Finance, 83, 57-69. Farrell, M. J. (1957). The measurement of productive efficiency. Journal of the Royal Statistical Society, 120(3): 253–281. Färe, R., & Grosskopf, S. (1985). A non-parametric cost approach to scale efficiency. The Scandinavian Journal of Economics, 594-604. Feinberg, R. M. (2001). The competitive role of credit unions in small local financial services markets. Review of Economics and Statistics, 83(3), 560-563. Fethi, M. D., & Pasiouras, F. (2010). Assessing bank efficiency and performance with operational research and artificial intelligence techniques: A survey. European journal of operational research, 204(2), 189-198. Ferguson, C & McKillop, D. G. (1997). The Strategic Development of Credit Unions, Chichester, John Wiley & Sons Limited, 231 pp. Fiordelisi, F., Marques-Ibanez, D., & Molyneux, P. (2011). Efficiency and risk in European banking. Journal of Banking & Finance, 35(5), 1315-1326. 196 University of Ghana http://ugspace.ug.edu.gh Flannery, M.J. (1974). An economic evaluation of credit unions in the United States, Federal Reserve Bank of Boston, Research Report No. 54. Frame, W. S., Karels, G. V., & McClatchey, C. A. (2003). Do credit unions use their tax advantage to benefit members? Evidence from a cost function. Review of Financial Economics, 12(1), 35-47. Frame, W.S. & White, L. (2010). Technological change, financial innovation, and diffusion in banking, in Berger, A., Molyneux, P. and Wilson, J.O.S. (eds.) Oxford Handbook of Banking. Oxford: Oxford University Press. Fried, H. O., Schmidt, S. S., & Lovell, C. K. (Eds.). (1993). The measurement of productive efficiency: techniques and applications. Oxford University Press. Froot, K. A., & Stein, J. C. (1998). Risk management, capital budgeting, and capital structure policy for financial institutions: an integrated approach. Journal of Financial Economics, 47(1), 55-82. Fujii, H., Managi, S., Matousek, R., & Rughoo, A. (2017). Bank efficiency, productivity, and convergence in EU countries: a weighted Russell directional distance model. The European Journal of Finance, 1-25. Fukuyama, H., Guerra, R., & Weber, W. L. (1999). Efficiency and ownership: evidence from Japanese credit cooperatives. Journal of Economics and Business, 51(6), 473-487. Gambacorta, L. (2005). Inside the bank lending channel. European Economic Review, 49, (issue number?) 1737-1759. Gambacorta, L., Scatigna, M., & Yang, J. (2014). Diversification and bank profitability: a nonlinear approach. Applied Economics Letters, 21(6), 438-441. 197 University of Ghana http://ugspace.ug.edu.gh Geoffrey M. Rubin George A. Overstreet Jr. Peter Beling, & Kanshukan Rajaratnam (2013). A dynamic theory of the credit union, Annals Operations Research (2013) 205:29–53. Glass, J. C., McKillop, D. G., Quinn, B., & Wilson, J. (2014). Cooperative bank efficiency in Japan: A parametric distance function analysis. The European Journal of Finance, 20(3), 291-317. Goddard, J., McKillop, D., & Wilson, J. O. (2008). The diversification and financial performance of US credit unions. Journal of Banking & Finance, 32(9), 1836-1849. Hannan, Timothy H.,(2003) The Impact of Credit Unions on the Rates Offered for Retail Deposits by Banks and Thrift Institutions. FEDS Working Paper No. 2003-06. Available at SSRN: http://dx.doi.org/10.2139/ssrn.386880 Isbister, J. (1994). Thin Cats: The Community Development Credit Union Movement in the United States. California: Center for Cooperatives, University of California. Jan Myers, Molly Scott Cato & Paul A. Jones (2012). An ‘alternative mainstream’? The impact of financial inclusion policy on credit unions in Wales, Public Money & Management, 32(6), 409-416. Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance, and takeovers. The American economic review, 76(2), 323-329. Jones, D. C., Kalmi, P., & Kauhanen, A. (2012). The effects of general and firm-specific training on wages and performance: Evidence from banking. Oxford Economic Papers, 64, 151-175. Hoff, A. (2007). Second stage DEA: Comparison of approaches for modelling the DEA score, European Journal of Operational Research 181 (2007) 425–435 198 University of Ghana http://ugspace.ug.edu.gh Kashyap, A.K., & Stein, J.C., (1994). Monetary policy and bank lending. In: Mankiw, N.G. (Ed.), Monetary Policy. Chicago University Press, Chicago, pp. 221–256. Koopmans, T.C., (1951). An analysis of production as an efficient combination of activities. In: Koopmans, T.C. (Ed.), Activity Analysis of Production and Allocation. Cowles Commission for Research in Economics, Monograph No. 13. Wiley, New York. Kumbhakar, S. C. & C. A. K. Lovell. (2000). Stochastic Frontier Analysis. Cambridge (England), New York: Cambridge University Press. Laeven, L., & Levine, R. (2007). Is there a diversification discount in financial conglomerates?. Journal of Financial Economics, 85(2), 331-367. Landskroner, Y., Ruthenberg, D., & Zaken, D., (2005). Diversification and performance in banking: The Israeli case. Journal of Financial Services Research, 27, 27–49. Lim, B., Lee, K., & Lee, C. (2016). Free Disposal Hull (FDH) Analysis for Efficiency Measurement: An Update to DEA. The Stata Journal. Mathuva, D. (2016). Revenue diversification and financial performance of savings and credit co-operatives in Kenya. Journal of Co-operative Organization and Management, 4(1), 1-12. Maudos, J., Pastor, J. M., Perez, F., & Quesada, J. (2002). Cost and profit efficiency in European banks. Journal of International Financial Markets, Institutions and Money, 12(1), 33-58. Mavenga, F. (2010). Economic Impact of Credit Unions on Rural Communities. University of Saskatchewan.10.1.1.857.3835. McKillop, D., C. Ferguson, & G. O'Rourke, (1997). "Typology for Credit Unions." ICA Review, Vol. 90 No. 1. 39-47. 199 University of Ghana http://ugspace.ug.edu.gh McKillop, D. G., Colin Glass, J., & Ward, A. M. (2005). Cost efficiency, environmental influences and UK credit unions, 1991 to 2001. Managerial Finance, 31(11), 72-86. McKillop, D. G., Glass, J. C., & Ferguson, C. (2002). Investigating the cost performance of UK credit unions using radial and non-radial efficiency measures. Journal of Banking & Finance, 26(8), 1563-1591. McKillop, D., & Wilson, J. O. (2011). Credit unions: A theoretical and empirical overview. Financial Markets, Institutions & Instruments, 20(3), 79-123. McKillop, D.G., Ward, A.M. & Wilson, J.O.S. (2011). Credit unions in Great Britain: Recent trends and current prospects', Public Money and Management, 31, 35-42. Mercieca, S., Schaeck, K., & Wolfe, S. (2007). Small European banks: Benefits from diversification? Journal of Banking & Finance, 31(7), 1975-1998. Meslier, C., Tacneng, R., & Tarazi, A. (2014). Is bank income diversification beneficial? Evidence from an emerging economy. Journal of International Financial Markets, Institutions and Money, 31, 97-126. Montgomery, C. A. (1994). Corporate Diversification. The Journal of Economic Perspectives. Vol. 8, pp. 163. Moody, J. C., & Fine, G. C. (1971). The Credit Union Movement. Origins and Development (Second). Credit Unions National Association. Mueller, D. (1969). A Theory of Conglomerate Mergers. Quarterly Journal of Economics. Vol. 83, pp. 643. Nguyen, T. P. T., Nguyen, T. P. T., Nghiem, S. H., & Nghiem, S. H. (2016). Market concentration, diversification and bank performance in China and India: An application of the two-stage approach with double bootstrap. Managerial Finance, 42(10), 980-998. 200 University of Ghana http://ugspace.ug.edu.gh Ofei, K A (2001). Retooling Credit Union: The Case of Credit Union Association of Ghana, IFLIP Research Paper, 01-3, ILO. Petersen, M., & Rajan, R., (1995). The effect of credit market competition on lending relationships. Quarterly Journal of Economics. 110, 407–443. Pitts, R. A., & Hopkins, H. D. (1982). Firm diversity: Conceptualization and measurement. Academy of management Review, 7(4), 620-629. Pruteanu‐Podpiera, A. M. (2007). The role of banks in the Czech monetary policy transmission mechanism. Economics of Transition, 15(2), 393-428. Quagliariello, M. (2007). Banks’ riskiness over the business cycle: a panel analysis on Italian intermediaries. Applied Financial Economics, 17(2), 119-138. Rajan, R. G., (1992). Insiders and outsiders: The choice between informed and arm's‐length debt. Journal of Finance, 47, 1367-1400. Rose, P. S. (1989). Diversification of the banking firm. Financial Review, 24(2), 251-280. Rosen, R. J., Lloyd-Davies, P. R., Kwast, M. L., & Humphrey, D. B. (1989). New banking powers: A portfolio analysis of bank investment in real estate. Journal of Banking & Finance, 13(3), 355-366. Rubin, G. M., Overstreet, G. A., Beling, P., & Rajaratnam, K. (2013). A dynamic theory of the credit union. Annals of Operations Research, 205(1), 29-53. Saghi-Zedek, N. (2016). Product diversification and bank performance: does ownership structure matter?. Journal of Banking & Finance, 71, 154-167. Sahoo, B. K., Mehdiloozad, M., & Tone, K. (2014). Cost, revenue and profit efficiency measurement in DEA: A directional distance function approach. European Journal of Operational Research, 237(3), 921-931. 201 University of Ghana http://ugspace.ug.edu.gh Sanya, S., & Wolfe, S. (2011). Can banks in emerging economies benefit from revenue diversification? Journal of Financial Services Research, 40(1-2), 79-101. Saunders, A., & Walter, I. (1994). Universal banking in the United States: What could we gain? What could we lose? Journal of Banking & Finance Vol 18, Issue 2, Pages 307- 323. Sealey, C. W., & Lindley, J. T. (1977). Inputs, outputs, and a theory of production and cost at depository financial institutions. The Journal of Finance, 32(4), 1251-1266. Servin R, Lensink R & Berg Marrit van den (2012). Ownership and technical efficiency of microfinance institutions: Empirical evidence from Latin America. Journal of Banking & Finance, 36 (4) 2136–2144. Sharpe, S.A. (1990). Asymmetric Information, Bank Lending and Implicit Contracts: A Stylized Model of Customer Relationships, Journal of Finance, 45(4), 1069.1087. Shephard, R.W., (1953). Cost and Production Functions. Princeton University Press, Princeton, NJ. Smith D, J (1988). Credit Union Rate and Earnings Retention Decisions under Uncertainty and Taxation, Journal of Money, Credit and Banking, 20(1), 119-131. Smith, D. J. (1984). A theoretic framework for the analysis of credit union decision making. The Journal of Finance, 39(4), 1155-1168. Smith, R, Staikouras, C & Wood, G(2003) Non-Interest Income and Total Income Stability. Bank of England Working Paper No. 198; Cass Business School Research Paper. Available at SSRN: http://dx.doi.org/10.2139/ssrn.530687 202 University of Ghana http://ugspace.ug.edu.gh Stein, J.C., (2002). Information production and capital allocation: decentralized vs. hierarchical firms. Journal of Finance, 57, 1891–1921. Stiroh, K. J. (2004). Diversification in banking: Is noninterest income the answer?. Journal of Money, Credit, and Banking, 36(5), 853-882. Stiroh, K. J., & Rumble, A. (2006). The dark side of diversification: The case of US financial holding companies. Journal of Banking & Finance, 30(8), 2131-2161. Sufian, F. (2009). Determinants of bank efficiency during unstable macroeconomic environment: Empirical evidence from Malaysia. Research in International Business and Finance, 23(1), 54-77. Taylor, R. A. (1977). Credit unions and economic efficiency. Rivista Internazionale Di Scienze Economiche E Commerciali, 24, 239–247. Templeton, W. K., & Severiens, J. T. (1992). The effect of nonbank diversification on bank holding company risk. Quarterly Journal of Business and Economics, 3-17. Tone, K. (2001). A slacks-based measure of efficiency in data envelopment analysis. European Journal of operational research, 130(3), 498-509. Tone, K. (2002). "A strange case of the cost and allocative efficiencies in DEA." Journal of the Operational Research Society 53(11): 1225-1231. Tone, K., & Sahoo, B. K. (2005). Evaluating cost efficiency and returns to scale in the Life Insurance Corporation of India using data envelopment analysis. Socio-Economic Planning Sciences, 39(4), 261-285. Tone, K., & Tsutsui, M. (2007). Decomposition of cost efficiency and its application to Japanese-US electric utility comparisons. Socio-Economic Planning Sciences, 41(2), 91-106. 203 University of Ghana http://ugspace.ug.edu.gh Tone, K., & Tsutsui, M. (2009). Network DEA: A slacks-based measure approach. European journal of operational research, 197(1), 243-252. Trinarningsih, W., Husa, P. P., Untoro, W., Trinugroho, I., & Sutaryo, S. (2016). Diversification and Bank Performance: Do TMT Characteristics Matter?. Advanced Science Letters, 22(5-6), 1651-1653. Trujillo‐Ponce, A. (2013). What determines the profitability of banks? Evidence from Spain. Accounting & Finance, 53(2), 561-586. Vihriälä, V. (1997). Banks and the Finnish Credit Cycle 1986–1995, Bank of Finland Studies E: 7. Helsinki: Bank of Finland. Wheelock, D. C., & Wilson, P. W. (2013). The evolution of cost-productivity and efficiency among US credit unions. Journal of Banking & Finance, 37(1), 75-88. Wijesiri M, Viganò L, & Michele, M (2015). Efficiency of microfinance institutions in Sri Lanka: a two-stage double bootstrap DEA approach. Economic Modelling, 47, 74–83. Wilcox, J. A., & Berkely, U. C. (2011). The increasing importance of credit unions in small business lending. Small Business Administration. Wilcox, J. A., & Dopico, L. G. (2011). Credit union mergers: Efficiencies and benefits. FRBSF Economic Letter, (vol & issue number?), 28. Williams, B. (2016). The impact of non-interest income on bank risk in Australia. Journal of Banking & Finance, 73, 16-37. Worthington, A. (2000). Cost efficiency in Australian non‐bank financial institutions: A non‐parametric approach. Accounting & Finance, 40(1), 75-98. Worthington, A. C. (1998). The determinants of non-bank financial institution efficiency: A stochastic cost frontier approach. Applied Financial Economics, 8(3), 279-287. 204 University of Ghana http://ugspace.ug.edu.gh Worthington, A. C. (1999). Measuring technical efficiency in Australian credit unions. The Manchester School, 67(2), 231-248. Zhou, K. (2014). The effect of income diversification on bank risk: evidence from China. Emerging Markets Finance and Trade, 50(sup3), 201-213. 205 University of Ghana http://ugspace.ug.edu.gh APPENDIX 1: NOTICE NO. BG/GOV/SEC/2017/06 OPERATING RULES AND GUIDELINES FOR CO-OPERATIVE CREDIT UNIONS AND CO-OPERATIVE FINANCIAL INSTITUTIONS (CFI) BANK OF GHANA NOTICE TO CO-OPERATIVE CREDIT UNIONS, CO-OPERATIVE FINANCIAL INSTITUTIONS AND THE GENERAL PUBLIC NOTICE NO. BG/GOV/SEC/2017/06 OPERATING RULES AND GUIDELINES FOR CO-OPERATIVE CREDIT UNIONS AND CO-OPERATIVE FINANCIAL INSTITUTIONS (CFI) In accordance with the Non-Bank Financial Institutions Act, 2008 (Act 774) and Co-operative Credit Union Regulations, 2015 (L.I. 2225), the following Operating Rules and Guidelines are hereby issued by the Bank of Ghana for the information and guidance of the general public and for strict compliance by entities operating Co-operative Credit Unions (Credit Union)/Co-operative Financial Institutions (CFI) in Ghana: 1. Regulated Activity Co-operative Credit Unions/CFI’s engaged in the activities that involve deposit taking, provision of credits and other financial services to their members constitutes regulated activity under Act 774. Except where expressly exempted in writing by the Bank of Ghana, Co-operative Credit Unions/CFI’s undertaking such activity are required to be registered and incorporated under sections 5 and 6 of the Co-operative Societies Act, 1968 (N.L.C.D. 252) and shall obtain a licence issued by the Bank of Ghana before commencing or continuing such activities. 2. Regulatory Requirements i. Business Form: the word “co-operative” shall form part of the name of every registered society. The word “limited” shall be the last word in the name of every registered society with limited liability. ii. Capital: Minimum share capital of every member shall not be less than one hundred Ghana Cedis (GHS 100) or such amount as may be determined by the Bank of Ghana. iii. Shareholding: No member other than a registered Society or Credit Union/CFI shall hold more than one-fifth of the share capital of another Co-operative Credit Union or Society. iv. Membership: A Co-operative Credit Union/CFI shall have a minimum membership of 150 with a potential membership of not less than 600. v. Branches and Agencies: Credit Unions/CFI shall not open, close or change a branch office, sub-office, agency or mobile 206 University of Ghana http://ugspace.ug.edu.gh unit in the country without the prior approval of the Bank of Ghana obtained at least thirty days before the date of opening, closing or change and having complied with any other conditions determined by the Bank of Ghana. vi. Amalgamation or Transfer of Societies or Credit Unions/CFI: Any two or more Societies or Credit Unions/CFI’s may by resolution passed by their respective general or special meetings and with the prior approval of the Bank of Ghana amalgamate as a single society. Amalgamation or transfer of assets and liabilities which involves the transfer of liabilities by one society to another society shall require three (3) months’ notice to be given to the creditors of the societies or society concerned who shall be entitled to a refund of any sum due them upon a written demand at one month before the date fixed for the amalgamation or transfer. The Bank may revoke the licence of a Credit Union/CFI in writing if it amalgamates or transfers its assets in a manner contrary to section 14 of NLCD 252. vii. Permissible activities: Credit Unions/CFI shall undertake the following permissible activities; a) Deposit taking: No single deposit shall exceed 10percent of the total deposit of the Co-operative Credit Union/CFI. A Credit Union/CFI wishing to exceed this limit shall obtain prior written approval of the Bank of Ghana. b) Granting of loans: A registered Society shall grant loans only to a member of that Society. However, a Society, with the sanction of a Central Society of which it is a member may grant a loan to another Society which is a member of the same Central Society. Loans granted to a member shall not be more than 10percent of the assets of the Credit Union/CFI. A Credit Union/CFI shall not grant a loan on the security of its own shares c) Investment: A registered Society or Credit Union/CFI may invest part of its funds in: i. Deposits in the Co-operative Credit Unions Association; ii. Shares of the Co-operative Credit Unions Association; iii. Bank deposits; 207 University of Ghana http://ugspace.ug.edu.gh iv. Treasury bills and bonds issued by the Government of Ghana/Bank of Ghana securities or certificate. Aggregate investments of a Credit Union/CFI in investments from other Credit Unions shall not exceed 1percent of the assets of the Co-operative Credit Union/CFI. d) Reserve Funds: A Co-operative Credit Union/CFI shall: i. Establish and maintain a statutory reserve fund into which shall be paid at least 25percent of the annual profits of the Co-operative Credit Union/CFI; ii. Pay into the Central Finance Facility an amount equivalent to 5percent of the total assets of the Credit Union/CFI at the end of each financial year. e) Borrowing: A Credit Union/CFI shall not: i. borrow more than 40percent of the assets of the Credit Union/CFI; ii. grant a security interest in its property except with the prior written approval of the Co-operative Credit Unions Association. f) Dividends: A Credit Union/CFI shall not pay dividend in cash in excess of 50percent of its annual profit without the prior approval of the Credit Union Association (CUA). viii. Non-permissible activities: A Credit Union/CFI shall not undertake any of the following: a. Issue checking accounts. However, a medium to large Credit Union/CFI may, with the approval of the Bank of Ghana, issue checking accounts; b. Engage in foreign exchange business: and c. Engage in any trading activities or hold any stocks of ix. Display of licence: A Cooperative Credit Union/CFI shall at all times display its licence or copies of its licence, its name and a statement of the fact that it is licensed to carry on business in a conspicuous position on the premises at which it carries on business. 3. Prudential Oversight i. Prudential Reporting: The Credit Union Association of Ghana (CUA) shall collect and collate statistics on the operations of Credit Unions/CFIs and furnish this to the Bank of Ghana periodically as may be determined by the Bank. ii. Submission of Returns: Credit Unions/CFIs shall submit prudential reports on assets, liabilities, income and expenditure to CUA of varying periodicity as may be determined by the Bank of Ghana. 208 University of Ghana http://ugspace.ug.edu.gh iii. On-site Supervision: Credit Unions/CFIs may be subject to on-site supervision by CUA of such periodicity as may be determined by the Bank of Ghana. iv. Capital Adequacy Ratio: A minimum Capital Adequacy Ratio of 15percent shall be maintained by a Credit Union/CFI or such other ratio as may be determined by the Bank of Ghana. v. Secondary Reserve Ratio: A Secondary Reserve Ratio of 18percent shall be maintained by a Co-operative Credit Union/CFI or such other ratio as determined by the Bank of Ghana. vi. Operating Licence: The Operating licence of Credit Unions/CFI’s shall be subject to annual renewal upon satisfactory performance and payment of the appropriate licence renewal fee as determined by Bank of Ghana. 4. Submission and Audited Financial Statements: A Credit Union/CFI shall: i. Submit a copy of its audited financial Statements to CUA not later than three (3) calendar months after the end of its financial year. ii. Exhibit throughout the year at a conspicuous place in every office and branch a copy of last audited financial statements. 5. Prohibited Actions: A Credit Union/CFI shall not without the approval of the Bank of Ghana change its name as contained in its licence; or in furtherance of the business for which it is licensed, use or refer to itself by a name other than the name under which it is licensed or by an abbreviation of that name unless the abbreviation has been approved by the Bank. Licensing Requirements The licensing requirements for Co-operative Credit Unions/CFIs are attached to this Notice. Amendments or modifications to this Notice The Bank of Ghana reserves the right to amend, add to or delete any or all of these Operating Rules and Guidelines as it deems fit from time to time. (Sgd.) CAROLINE OTOO (MRS) THE SECRETARY March 8, 2017 209 University of Ghana http://ugspace.ug.edu.gh APPENDIX 2A. Non-Financial Income Diversification Histogram 0 .2 .4 .6 .8 1 hhinfdi HHINII  Figure 2.a : Non - interest income diversification index score APPENDIX 2B. Liquid-Financial Investment Diversification Histogram .2 .4 .6 .8 1 hhilfiii HHILFI  Figure 2.b: Liquid-Financial Investment diversification Index Score 210 Density Density 0 2 4 6 0 1 2 3 4 University of Ghana http://ugspace.ug.edu.gh APPENDIX 2.C: Combine Non-Loan Income Diversification Histogram .2 .4 .6 .8 1 coHmbHdiIvinc COMB Figure 3.c: Combine non - loan diversification Income Score 211 Density .5 1.5 0 1 2 University of Ghana http://ugspace.ug.edu.gh APPENDIX 3.A: Tone’s Efficiency Scores Histogram 0 .2 .4 .6 .8 1 toneeffi Tone’s Efficiency Figure 3.a.: Tone’s Efficiency Scores APPENDIX 3.B: Technical Efficiency Score Histogram 0 .2 .4 .6 .8 1 techeffici Technical Efficiency Figure 3.b: Histogram of Technical Efficiency Scores 212 Density Density .5 1.5 0 1 2 0 1 2 3 4 University of Ghana http://ugspace.ug.edu.gh APPENDIX 4a: Scatter plots 4 5 6 7 8 size toneeffi Fitted values 0 20 40 60 80 100 liqutotalass techeffici Fitted values 213 .2 .4 .6 .8 .2 .4 .6 .8 0 1 0 1 University of Ghana http://ugspace.ug.edu.gh APPENDIX 4a. Scatter plots 0 20 40 60 80 100 liqutotalass loantoassets Fitted values 4 5 6 7 8 size hhinfdi Fitted values 214 100 .2 .4 .6 .8 20 40 60 80 0 1 0 University of Ghana http://ugspace.ug.edu.gh APPENDIX 4a. Scatter plots 0 10 20 30 40 nieta hhilfiii Fitted values 0 20 40 60 80 100 loantoassets combdivinc Fitted values 215 .2 .4 .6 .8 .2 .4 .6 .8 1 1 University of Ghana http://ugspace.ug.edu.gh APPENDIX 5.a: Non- Constant Integrated Cost Efficiency Financial Performance Table 4.11a: Integrated Cost Efficiency Financial Performance. (1) (2) (3) (4) (5) (6) VARIABLES RAROE RAROA RAROE RAROA RAROE RAROA HHI  NII -0.8979*** -0.9134*** (-3.0219) (-3.0563) SIZA 0.1914* 0.1326 0.279 4** 0.221 8** 0.249 4** 0.19 60* (1.6723) (1.1520) (2.4967) (1.9707) (2.2119) (1.7262) L OTA 0.0131* 0.0124* 0.0165** 0.0158** 0.0154** 0.0149** (1.9065) (1.7889) (2.4028) (2.2825) (2.2478) (2.1559) N IM 0.0495*** 0.0497*** 0.0479*** 0.0482*** 0.0504*** 0.0503*** (8.0744) (8.0695) (7.6485) (7.6565) (7.9735) (7.9003) LITA 0.0141** 0.0135* 0.0232*** 0.0226*** 0.0202*** 0.0201*** (2.0387) (1.9348) (3.6182) (3.5036) (3.0997) (3.0649) L LP -0.0684*** -0.0663*** -0.0702*** -0.0682*** -0.0677*** -0.0659*** (-6.0295) (-5.8060) (-6.0952) (-5.8854) (-5.8920) (-5.6933) N WTA -0.0364*** -0.0310*** -0.0429*** -0.0377*** -0.0435*** -0.0382*** (-3.9333) (-3.3328) (-4.6920) (-4.1013) (-4.7928) (-4.1736) Z SCORE 3.7220*** 3.7941*** 3.7381*** 3.8104*** 3.7190*** 3.7937*** (24.2756) (24.6042) (24.1384) (24.4602) (24.0455) (24.3606) B KCN3 0.0108 0.0035 -0.0006 -0.0080 -0.0018 -0.0092 (0.5373) (0.1721) (-0.0295) (-0.3996) (-0.0907) (-0.4585) I NFL -0.0126 -0.0106 -0.0091 -0.0072 -0.0093 -0.0072 (-0.5475) (-0.4569) (-0.3900) (-0.3064) (-0.4036) (-0.3085) GDP -0.0332 -0.0308 -0.0400 -0.0380 -0.0354 -0.0337 (-1.1986) (-1.1083) (-1.4298) (-1.3502) (-1.2682) (-1.2005) CE 2.0285*** 2.1453*** 1.6973*** 1.8045*** 1.8046*** 1.9027*** (5.4842) (5.7667) (4.7185) (4.9868) (5.0011) (5.2367) HHI LFI  0.0466 0.0847 (0.1239) (0.2236) HHI  COMB 0.5555* 0.4845 (1.7796) (1.5416) C redit Unions 61 6 1 6 1 61 61 61 Observations 427 427 427 427 427 427 Wald chi2(12) 1888.89 1881.08 1840.48 1831.93 1857.22 1844.24 Prob chi2 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 z-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 216 University of Ghana http://ugspace.ug.edu.gh APPENDIX 5.b. Non-Constant Integrated Technical Efficiency Financial Performance Table 4.11a: Integrated Cost Efficiency Financial Performance. (1) (2) (3) (4) (5) (6) VARIABLES RAROE RAROA RAROE RAROA RAROE RAROA HHI NFI  -0.5582** -0.5470** (-2.4920) (-2.4097) S IZE 0.3422*** 0.3024*** 0.3686*** 0.3277*** 0.3687*** 0.3291*** (4.3463) (3.7907) (4.6886) (4.1156) (4.6930) (4.1356) COTI -0.0280*** -0.0281*** -0.0279*** -0.0280*** -0.0279*** -0.0280*** (-17.4761) (-17.3120) (-17.3016) (-17.1484) (-17.2919) (-17.1451) L OTA 0.0103* 0.0098* 0.0124** 0.0118** 0.0124** 0.0119** (1.8801) (1.7609) (2.2646) (2.1240) (2.2610) (2.1465) N IM 0.0228*** 0.0226*** 0.0227*** 0.0226*** 0.0228*** 0.0222*** (4.7068) (4.5927) (4.5629) (4.4866) (4.5828) (4.4057) L ITA 0.0125** 0.0126** 0.0176*** 0.0175*** 0.0175*** 0.0177*** (2.1707) (2.1602) (3.2322) (3.1715) (3.2125) (3.2156) L LP -0.0053 -0.0038 -0.0052 -0.0039 -0.0049 -0.0035 (-0.5471) (-0.3908) (-0.5276) (-0.3889) (-0.4984) (-0.3559) NWTA -0.0164** -0.0116 -0.0200*** -0.0153** -0.0201*** -0.0149** (-2.2140) (-1.5451) (-2.7105) (-2.0426) (-2.7258) (-1.9987) ZSCORE 3.5308*** 3.6013*** 3.5438*** 3.6142*** 3.5405*** 3.6140*** (29.2291) (29.4240) (29.1538) (29.3618) (29.0534) (29.2785) BKCN3 0.0169 0.0084 0.0117 0.0034 0.0111 0.0031 (1.0890) (0.5320) (0.7524) (0.2156) (0.7167) (0.1942) I NFL -0.0048 -0.0024 -0.0030 -0.0009 -0.0027 -0.0003 (-0.2638) (-0.1289) (-0.1647) (-0.0467) (-0.1464) (-0.0186) G DP -0.0082 -0.0055 -0.0134 -0.0109 -0.0122 -0.0103 (-0.3749) (-0.2498) (-0.6118) (-0.4932) (-0.5579) (-0.4626) TE 0.4335* 0.4369* 0.3785 0.3769 0.3721 0.3905* (1.9090) (1.8986) (1.6418) (1.6147) (1.6088) (1.6670) HHI  LFI 0.0656 0.1158 (0.2214) (0.3862) HHI  COMB 0.0878 -0.0007 (0.3572) (-0.0029) Credit Union 61 61 61 61 61 61 Observations 427 427 427 427 427 427 Wald chi2 3329.29 3261.52 3275.87 3213.30 3276.55 3212.03 Prob chi2 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 z-statistics in parentheses*** p<0.01, ** p<0.05, * p<0.1 217 University of Ghana http://ugspace.ug.edu.gh APPENDIX 6A. Sample Comprehensive Income of a Credit Union 218 University of Ghana http://ugspace.ug.edu.gh APPENDIX 6b. Sample Statement of Financial Position of a Credit Union 219