Michael Nyaga

MACROECONOMIC FACTORS AFFECTING LOAN PERFORMANCE OF COMMERCIAL BANKS IN KENYA MICHAEL GITARI NYAGA A research project report submitted to the Department of Economics, Accounting and Finance in the School of business in partial fulfillment of the requirements for the award of the degree of Post Graduate Diploma in Business Studies Jomo Kenyatta University of Agriculture and Technology 2015 DECLARATION This research project report is my own original work and has not been presented for examination in any other University. _____________________ _____________________ Signature Date Michael Gitari Nyaga Student No: HDB-312-040-2204/2014 This research project report has been submitted for examination with my approval as the University Supervisor. ____________________ ____________________ Signature Date Dr. Willy Muturi JKUAT, KENYA ii DEDICATION I dedicate this research project report to my Dad, Mr. Christopher Nyaga, Mom, Mrs. Peris Nyaga, my siblings, Sammy, Henry and Victor and to Steve Kelly, my mentor. iii ACKNOWLEGEMENTS Though I will not be able to list all those who were of help to me during the process of my studies and research, I would like to acknowledge with appreciation, the following people who made this research possible, but above all, I thank the Almighty God for His favor and blessings throughout my studies, and in my entire life. Thank you Lord. Dr. Margaret Ncabira, my research methods professor, who provided great insight and guidance in class. It was a great learning experience that will enhance by technical capabilities in my professional life. Dr. Willy Muturi, my research supervisor for being so tolerant and understanding, while being so helpful in enabling me to complete my research proposal and subsequent defense. Dr. Benjamin Maturu: at the research center, who allowed for any questions and also gave sufficient consul. Dr. Tobias Olweny, my corrections defense panel supervisor and former professor (Portfolio Management Class), for his friendly but serious approach to research application. Dr. Moses Kiptui: head of the research center, who took every opportunity to help with any research complications iv TABLE OF CONTENTS Contents .............................................................................................................................................i DECLARATION .............................................................................................................. ii DEDICATION ................................................................................................................. iii ACKNOWLEGEMENTS .................................................................................................iv TABLE OF CONTENTS ................................................................................................... v LIST OF TABLES ......................................................................................................... viii LIST OF FIGURES ..........................................................................................................ix ABREVIATIONS .............................................................................................................. x ABSTRACT ......................................................................................................................xi CHAPTER ONE: INTRODUCTION ................................................................................ 1 1.1 Background of the Study.............................................................................................. 1 1.2 Statement of the Problem ............................................................................................. 6 1.3 Research Objectives ..................................................................................................... 9 1.3.1 Objectives-General .................................................................................................... 9 1.4 Research questions ....................................................................................................... 9 1.5 Justification of the study .............................................................................................. 9 1.6 Scope of the study ...................................................................................................... 10 1.7 Limitation of the study ............................................................................................... 11 CHAPTER TWO: LITERATURE REVIEW .................................................................. 12 2.1 Introduction ................................................................................................................ 12 2.2 Theoretical Literature ................................................................................................. 12 2.2.1 Theory of asymmetric information ......................................................................... 12 v 2.2.2 Adverse Selection Theory ....................................................................................... 13 2.2.3 Moral Hazard Theory .............................................................................................. 13 2.2.4 Agency Theory ........................................................................................................ 14 2.2.5 Financial Theory ..................................................................................................... 15 2.3 Conceptual Framework .............................................................................................. 16 2.4 Empirical Review ....................................................................................................... 18 2.4.1 Interest rates ............................................................................................................ 19 2.4.2 Real GDP Growth Rate ........................................................................................... 20 2.4.3 Inflation Rate ........................................................................................................... 22 CHAPTER THREE: RESEARCH METHODOLOGY .................................................. 24 3.1 Introduction ................................................................................................................ 24 3.2 Research Design ......................................................................................................... 24 3.3 Target Population and Sampling. ............................................................................... 25 3.4 Data Collection Procedures ........................................................................................ 25 3.5 Data Processing and Analysis .................................................................................... 26 3.6 Model Specification (Time series model) .................................................................. 27 CHAPTER FOUR: RESEARCH RESULTS AND DISCUSSIONS .............................. 29 4.1 Introduction ................................................................................................................ 29 4.2 Univariate Analysis .................................................................................................... 29 4.3 Correlation Analysis................................................................................................... 31 4.4 Stationarity Test ......................................................................................................... 32 4.5: Cointegration Analysis .............................................................................................. 35 4.6 Vector error correction model analysis ...................................................................... 36 4.7 Discussion of residual diagnostic tests ....................................................................... 38 4.8: Results Discussion .................................................................................................... 39 CHAPTER FIVE: SUMMARY, CONCLUSIONS AND RECOMMENDATIONS ..... 42 5.1 Introduction ................................................................................................................ 42 vi 5.1 Summary .................................................................................................................... 42 5.3 Conclusions ................................................................................................................ 42 5.4 Recommendations ...................................................................................................... 43 5.5 Suggestions for Further Research .............................................................................. 44 References ........................................................................................................................ 45 APPENDIX 1: STATIONALITY GRAPHS ................................................................... 51 APPENDIX II Sample Data Collected And Transformed (1+r) ...................................... 52 vii LIST OF TABLES Table 2.1: Definition of variables under study………………………………………….14 Table 4.1: Descriptive statistics…………………………………………………………25 Table 4.2: Correlation Table…………………………………………………………….26 Table 4.3: Stationality test (Augmented Dickey-Fuller)………………………………..27 Table 4.4: Stationality test (Phillips Perron)……………………………………………28 Table 4.5: Johansen cointegration test………………………………………………….29 Table 4.6: Vector error correction model……………………………………………….30 Table 4.7: Summary of residual diagnostic tests………………………………………..32 Table 4.8: OLS model equation estimate……………………………………………….39 viii LIST OF FIGURES Figure 2.1: Conceptual framework……………………………………………………14 ix ABREVIATIONS NPLs Non-Performing Loans CBK Central Bank of Kenya KCB Kenya Commercial Bank NBK National Bank of Kenya MPC Monetary Policy Committee KBRR Kenya Bank Reference Rate CBR Central Bank Rate APR Annual Percentage Rate GCC Gulf Co-operation Council x ABSTRACT This study sought to establish the factors that affect loan performance of Commercial Banks in Kenya. Commercial Banks in Kenya experience high levels of NPLs a trend that threatens the viability and sustainability of Banks and hinders the achievements of their goals. The subject matter therefore dwelt on non-performing loans (NPLs) to total loans as the sole dependent variable. The independent variables were solely macroeconomic indicators, and they included; Interest rates, Real GDP Growth Rate and Inflation Rate. The macroeconomic environment was viewed as a critical driver for nonperforming loans. In this regard, the main goal of this study was to emphasize the relationships of the above macroeconomic independent variables to the sole dependent variable, nonperforming loans to total loans. The period covered under this study was from 2008 to 2015 third quarter, where monthly secondary data was used. The population of the study comprised of all 43 commercial banks in Kenya. The study used time series data to model the relationships between non-performing loans and the selected macroeconomic variables. xi CHAPTER ONE: INTRODUCTION 1.1 Background of the Study Non-performing loans (NPLs) refers to those financial assets from which banks no longer receive interest and or installment payments as scheduled. In short, they cease to “perform” or generate income for that particular financial institution. Hennie (2003) describes NPLs as those loans which are not generating income. While (Bexley and Nenninger, 2012) refer NPLs as those loans that are ninety or more days delinquent in payments of interest and or principal. Alton and Hazen (2001) concur, and describe NPLs as loans which have been ninety days or more past due and no longer accruing interest. Choudhury et al (2002: 21-54) classified NPLs as a “multiclass” concept, based on the length of overdue of the said loan(s). According to Woo (2000), NPLs should be viewed as a typical byproduct of an impending financial crisis, in that, they bring down investors confidence in the banking system. They have been identified as one of the main reasons that cause insolvency of the financial institutions and the entire economy (Hou, 2007). In the U.S, studies show that the level of NPLs as early as 2006 had already started to spiral out of control across all financial institutions prior to the collapse of the sub-prime mortgage market in August 2007 (Greenidge and Grosvenor, 2010). Ahmad (2002) while analyzing the Malaysian financial system concluded that there was a significant relationship between NPLs and financial crisis and went further to assert that NPLs had already started to build up before the onset of the 1997 Asian financial crisis. Nishimura et al (2001) studied the financial situation in Japan and opined that a significant amount of loans disbursed to firms during the bubble era ended up being non-performing when the inevitable bubble went burst. Caprio and Klingebiel (2002) concluded that NPLs in Indonesia represent about 75 percent of total loan assets, which ultimately led to the collapse of over sixty banks in the 1997 Asian financial crisis. In the Sub-Saharan African countries in the 1990s, the banking crisis that engulfed the region was also accompanied by a rapid accumulation of NPLs (Fofack, 2005). While in East Africa, commercial banks are feeling the weight of rising NPLs brought about by a regime of high interest rates coupled by stringent local Central Banks regulatory requirements. Banking data released by various national banks and regulators reveal that the volume of NPLs has been rising since 2012 (CBK Annual Report, 2013). In Uganda, NPLs rose 40 percent in the last year ending December 2013 due to a volatile economic environment. Growth in NPLs in Uganda mirrors that of Kenya where NPLs grew by 30 percent to Ksh.80 billion in the same period (CBK Annual Report, 2014). In Kenya, the 1985-1986 and 1990-1998 era, brought with it a banking crisis that ultimately culminated in major banks going under, thirty seven failed banks as at 1998. The banks failures were mainly attributed to high NPLs primarily enabled by high interest rate spreads (Kithinji and Waweru, 2007). According to CBK Supervision Report 2013, non-performing loans(NPLs) in Kenya`s banking sector shot up by 13.33% 2 in 2012 due to the prevalence of high interest rates, while the gross non-performing loans (hereafter NPLs) rose to $716.25 million as at end of December 2012. A lot of studies have been conducted on the issue of nonperforming loans in the banking sector. For instance, the study of Calice (2012) for the Tunisian banking sector, Blanco and Gimeno (2010) for South African banks and Kolapo (2012) for the Nigerian banks. The studies showed that NPLs have an adverse effect on banking sectors survival, and as such, the cause for NPLs should be given due consideration. Its causes are different in each country of study and that might be due to situational factors. Accordingly, this issue has attracted the interest of different researchers in different countries. That means a lot of studies have been performed on the determinants of NPLs of financial sectors worldwide. The relationship between the macroeconomic variables and NPLs has also attracted attention in literature (see for example Hoque and Hossain,2004; Siddiqui, Malik and Shah,2011). Among factors cited by the literature as significant determinants include: the Interest rate, the annual Real Gross Domestic Product (hereafter GDP) Growth, the annual Inflation Rate, loans growth, the Unemployment Rate, the Money Supply (M2), the Real Effective Exchange Rate, and Index of Production (Adebola, Yusoff and Dahalan, 2011; Asari et al., 2011; Rinaldi and Sanchis-Arellano, 2006; Salas and Saurina, 2002). These macroeconomic variables affect economic conditions on households and firms and as a result, influence their ability to repay the loans. When the 3 economy is expanding, there is a relatively few number of NPLs, as both consumers and firms have enough returns to repay their debts. However, as boom time continues, granting of loans is extended to less credit worthy borrowers and later, when the recession time starts, the level of NPLs increases (Quagliarello, 2007). Further, recessionary periods are often associated with low GDP. When the GDP shrinks, households and firms end up having reduced streams of income and consequently higher levels of NPLs (Salas and Saurina, 2002). To support this notion, Basel (2013) using estimation technique method and panel data set Covering 75 countries over ten year period from 2005 to 2010 studied the macroeconomic determinants of non-performing loans. The Analysis presented that real GDP growth was the main driver of nonperforming loan ratio. Some literature reviewed, has proved that inflation rate and NPLs are positively related. Fofack (2005) showed that inflation contributed to the increase of bad loans in most of African countries. Further, he showed that inflation was evident in the rapid erosion of commercial banks` equity and thus higher NPLs in those banks. However, inflation affects borrowers’ debt servicing capacity through different channels and its impact on NPL can be positive or negative. Higher inflation can make debt servicing easier for the borrower by reducing the real value of outstanding loans. However, it can also weaken the borrowers’ ability to service debt by reducing real income when wages are sticky. Moreover, when loan rates are variable (floating), 4 inflation is likely to reduce borrowers’ loan servicing capacity as lenders adjust rates to maintain their real returns or simply to pass on increases in policy rates resulting from monetary policy actions to combat inflation. Against this background, the relationship between NPL and inflation can be positive or negative. In addition, literature on macroeconomic determinants of NPLs has also been sufficiently covered in most of the developed and emerging economies. For instance, Saba et al.(2012) studied on the determinants of NPLs on US Banking sector and found that lending rates had negative while Inflation and Real GDP per capital had positive and significant effect on NPLs. Louzis et al.(2010) examined the determinants of NPLs in the Greek financial sector using dynamic panel data model and found Real GDP growth rate, ROA and ROE had negative whereas lending rates, unemployment and inflation rate had positive significant effect on NPLs. The study of Skarica (2013) on the determinants of NPLs in Central and Eastern European Countries (CEEC), through fixed effect model found Real GDP growth rate, unemployment rate and inflation had negative and significant impact on NPLs. Similarly, Carlos (2012), based on Ordinary Least Squares (OLS) model estimators, found that NPLs have negative association with GDP growth rate whereas a positive association with unemployment rate. Karumba and Wafula (2012) in their article on alternatives for Kenyan banking industry, identified non-performing loans as being one of the oldest and most challenging risk(s) faced by banks, which results due to the probability that borrowers may default. 5 Given the recent turbulence in banking and the rise in non-performing loans (NPLs) there is renewed interest in the impact of internal and external factors on NPLs of banks. Most of the research studies done in Kenya have largely dwelt on interest rate spread and its impact on NPLs, or credit risk management of commercial banks in Kenya or both bank specific and macroeconomic determinants affecting non-performing loans of commercial banks in Kenya. For instance: Collins, N.J (2011), on the effects of interest rate spread of non-performing assets; Muriithi, M.W(2013), on the cause of NPLs in the commercial banks in Kenya; Ongore and Kusa(2013), on the determinants of financial performance of commercial banks in Kenya; Warue, B.N(2013), on the effects of bank specific and macroeconomic factors on NPLs in the commercial banks in Kenya. It is in this context that the study seeks to concentrate solely on the macroeconomic factors affecting loan performance of commercial banks in Kenya. The studies independent macroeconomic variables will include: Interest rates, Real GDP Growth Rate and Inflation Rate, while the dependent variable will be non-performing loans. 1.2 Statement of the Problem The purpose of the study was to investigate the factors that affect loan performance of Commercial Banks in Kenya. For any financial institution, loan performance in regards to NPLs is a key area of study. A high provision of NPLs adversely impacts on profitability and performance of total assets. As such, controlling NPLs is extremely crucial for both the performance aspect of an individual bank and the entire financial 6 system (McNulty, Akhigbe, and Verbrugge, 2001). According to Kroszner (2002), NPLs are closely associated with banking crisis. NPLs should then be treated as undesirable outputs or costs to a loaning bank, which in turn decreases the banks performance (Chang, 1999). According to the Central Bank of Kenya (CBK) Annual Report 2014, total NPLs in the banking industry 2013, jumped by 30.9% to Kenya shillings, 80.6 billion. One of the major factors implicated was due to high interest rates in 2012. In an effort to contain deteriorating levels of NPLs, the Monetary Policy Committee (MPC) on July 8th, 2014, launched the Kenya Bank Reference Rate (KBRR). The new reference rate (base rate) is expected to guide lending across the banking industry and standardize the way lending rates are set. The main objective is to infuse transparency in the pricing of bank loans and that credit is affordable and accessible in the market (CBK Annual Report, 2014). It is calculated based on the Monetary Policy rate or Central Bank rate (CBR) average and the 91-day Treasury bill average yield over six months. Of the total NPLs in the banking industry of 80.6 billion, Central Bank supervision report (2013) indicated that the personal and household sector constituted 26% of the total NPLs followed closely by the trade sector at 24.7%. This as a result signaled an increase in credit risk in the overall financial sector. In May 2014, banks in Kenya adopted the Annual Percentage Rate (APR), to enable customers to compare loan costs based on a standard computation model. Since the APR discloses the entire cost of a loan, customers are able to shop for 7 the best rate around (CBK Supervision Report, 2014). NPLs levels have also dramatically increased in the past year due to the new prudential guidelines enacted by the Central Bank of Kenya, in their 2013 credit report survey. The new guidelines require lenders to classify as non-performing accounts of a borrower who defaults on the repayment of any one of multiple loans for more than three months. Put together, all this has added to higher levels of credit risk and undue moral hazard on the part of the borrower. Previous studies on the Kenyan context concentrated largely on the effects of credit risk management on performance of commercial banks in Kenya. This includes: Kithinji and Waweru (2007), on credit risk management and profitability of commercial banks in Kenya; Warue (2013), on the effects of bank Specific and macroeconomic factors on nonperforming Loans in Commercial Banks in Kenya; A Comparative Panel Data analysis, Musyoki (2011) and Ogilo (2012), who separately conducted an empirical study on the impact of credit risk management on financial performance of Kenyan banks and Ngetich (2011), who analyzed the effects of interest rates spread on the level of non-performing assets on commercial banks in Kenya. A close scrutiny of these studies shows that the evidence on macroeconomic determinants of NPLs in commercial banks in Kenya is scant. This study aimed at filling this research gap. 8 1.3 Research Objectives 1.3.1 Objectives-General The main objective of this study was to investigate the macroeconomic factors affecting loan performance of Commercial Banks in Kenya. Specifically the study sought to Identify: 1. To establish the effects of Interest rates on loan performance. 2. To evaluate the effects of Real GDP Growth Rate on loan performance. 3. To determine how the Inflation Rate affects loan performance. 1.4 Research questions Consistent with the research objectives outlined in section 1.3.1, the study sought answers on the following research questions. 1. To what extent do Interest rates influence loan performance? 2. Does Real GDP Growth Rate affect loan performance? 3. What is the impact of Inflation Rate on loan performance? 1.5 Justification of the study The research findings of this study helped in addressing the existing knowledge gap in literature of effects of macroeconomic variables on NPLs in Kenyan commercial banks. It was also be a valuable addition to the existing knowledge and provided a platform for further research which was useful to scholars. An understanding of the effects of the 9 macroeconomic variables on NPLs in the Kenyan banking system was important to the senior management and investors of financial institutions in Kenya. The study findings enabled managers and investors to make timely decisions on how to avoid risk, transfer risks, risk reduction (mitigating risk) or retain the risk in a bid to maximize returns. On the policy front, the study findings were important to the government, regulatory bodies and to the commercial banks themselves. It helped the regulators to know exactly how NPLs were affected by macroeconomic variables and how to strengthen the financial industry in terms of policies. 1.6 Scope of the study This report was divided into five chapters. In the first chapter, the background of the selected research area was presented. In chapter two, the literature review of studies related to the topic was outlined. Chapter three presented the research Methodology which detailed how the correlation matrix, OLS estimate and regression analysis, were used to link non-performing loan ratio and its determinants. Chapter four presented the study findings of the research project. It clearly showed that macroeconomic factors (Interest rate and Real GDP growth rate) were critical in explaining non-performing loan ratio in a financial institution. Lastly chapter five was about data analysis. It outlined results and interpretations of the OLS model estimate and the correlation matrix and ended with conclusion of the study. 10 1.7 Limitation of the study The study was limited to a small number of observations (sample) due to the unavailability of concise and reliable research data. However, the study sought to remedy this with monthly data up to the present third quarter (2015) data inputs, hence increasing the sample size moderately. Also, the data had to be transformed (1+r) format to eliminate the presence of outlier inputs and to improve tests results. 11 CHAPTER TWO: LITERATURE REVIEW 2.1 Introduction This chapter covers a literature review relevant to the study. It contains theories related to non-performing loans, the dependent variable. The chapter covered empirical data pertinent to the macroeconomic independent variables namely, Interest rates, Real GDP Growth Rate and Inflation Rate. 2.2 Theoretical Literature This study limited its focus on competing theories relevant primarily to, non-performing loans, the dependent variable. There are five theories that have provided insight into how these factors influence nonperforming loans levels. 2.2.1 Theory of asymmetric information Asymmetry theory proposes a situation where it may be difficult to distinguish between good and bad borrower, which may result into adverse selection and moral hazard problems (Auronen, 2003). It`s based on a situation where a party does not have sufficient information about the other party to make an accurate investment decision. Akerlof (1970) was the first to identify potential inefficiencies created by information asymmetry in financial markets. He opined that due to lack of sufficient information, a lender might issue credit to an undesirable borrower (adverse selection), or the borrower might engage in activities that are contrary to the initial terms and conditions of the loan 12 contract (moral hazard). Studies show that adverse selection and moral hazard contribute to significant accumulation of NPLs in the banking system (Bester, 1994; Bofondi and Gobbi, 2003). Seminal contributions on asymmetric information theory also include Spence (1973), Weiss (1983), Erik Bond (1982), Grossman and Hart (1983) and Rothschild and Stiglitz (1976). 2.2.2 Adverse Selection Theory The adverse selection problem states that when lenders cannot differentiate between good from bad borrowers, then all borrowers risk profiles are merged to reflect their pooled market risk profile and charged a risk adjusted interest rate (Auronen, 2003). If this rate is considerably higher than good borrowers rationalize, then, it will price them out of the market. As a result, the remaining borrowers will incur even higher interest rates (Akerlof, 1970). Pagano and Jappelli (1993) opine that information sharing between different banks reduces the problem of adverse selection, and distinguishes good clients from bad ones. Good borrowers with low risk profiles would receive more attractive interest rates while high risk borrowers would be rationed out of the market (Barron and Staten, 2008). Additional contributions on this topic, has also been covered by Boyan Jovanovic (1982), Hayne Leland (1979) and Joseph Stiglitz (1975). 2.2.3 Moral Hazard Theory The moral hazard problem states that a borrower has a high probability of defaulting unless such action(s) reflects on his or her future acquisition of credit. This is due to the difficulty lenders have in assessing the actual level of wealth the borrower will have 13 accumulated prior to the date the debt is due as opposed to the moment of credit application. If the lenders cannot assess the borrower`s wealth then the latter will be in a position to default on the loan. To mitigate this risk, lenders will be inclined to increase interest rates charged on loans leading to the eventual breakdown of the market (Alary and Goller, 2001). Additional contributions on this topic can be seen in the seminal papers of Hart and Holmstrom (1987), Salanie (1997), Stadler and Castrillo (1997), and Hermalin et al. (1991). 2.2.4 Agency Theory According to the Agency theory, the principal agency problem can be reduced by better monitoring such as establishing more appropriate incentives for managers. In the field of corporate risk management agency issue have been shown to influence managerial attitudes towards risk taking and hedging Smith and Stulz(1985). Theory also explains a possible mismatch of interest between shareholder management and debt holders due to asymmetries in earning distribution, which can result in the firm taking too much risk or not engaging in positive net value project (Smith and Stulz, 1987). Consequently, agency theory implies that defined hedging policies can have important influence on firm value (Fite and Pfleiderer, 1995). As stated by Lambert (2001), agency theory evaluates the impact of the conflict of interest between principals and agents because of (1) shirking by the agent, (2) diversion of resources by the agent for private consumption, (3) differential time horizon of the agent and principal, (4) differential risk 14 aversion of the agent and the principal. Additional seminal contributions include Shavell and Steven (1970), Sappington and David (1983), and Rogerson and William (1985). 2.2.5 Financial Theory Financial theory pioneered by, Minsky (1974) also known as financial instability hypothesis, attempted to provide an understanding and explanation of the characteristics of financial crisis. The theory suggests that, in prosperous times, when corporate cash flow rises beyond what is needed to pay off debt, a speculative euphoria develops, and soon thereafter debts exceed what borrowers can pay off from their incoming revenues, which in turn produces a financial crisis. As a result of such speculative borrowing bubbles, banks and lenders tighten credit availability, even to companies that can afford loans and the economy subsequently contracts. The theory identifies three types of borrowers that contribute to the accumulation of insolvent debt: The "hedge borrower" can make debt payments (covering interest and principal) from current cash flows from investments. For the "speculative borrower", the cash flow from investments can service the debt, i.e., cover the interest due, but the borrower must regularly roll over, or reborrow, the principal. The "Ponzi borrower" borrows based on the belief that the appreciation of the value of the asset will be sufficient to refinance the debt but cannot make sufficient payments on interest or principal with the cash flow from investments; only the appreciating asset value can keep the Ponzi borrower afloat. Financial theory underpin this study in that, a hedge borrower would have a normal loan and is paying 15 back both the principal and interest; the speculative borrower would have a watch loan; meaning loans’ principal or interest is due and unpaid for 30 to 90 or have been refinanced, or rolled-over into a new loan; and the Ponzi borrower would have a substandard loan, meaning the payments do not cover the interest amount and the principal is actually increasing. The primary sources of repayment are not sufficient to service the loan. The loan is past due for more than 90 days but less than 180 days. Substandard loans are nonperforming loans, hence applicability of financial theory in this study. Similar seminal contributions on this topic can also been seen in the works of Irving Fisher (1933), Charles Kindleburger (1978), Martin Wolfson (1986), Joseph Schumpeter (1934) and Keynes (1972). 2.3 Conceptual Framework A conceptual framework is a visual or written product, one that explains either graphically or in narrative form, the main things to be studied-the key factors, concepts or variables, and the presumed relationships among them (Miles and Huberman, 1994). 16 NPL == Outstanding principal balance of loans past due more than (90) days Outstanding principal balance of all loans INTEREST RATES REAL GDP GROWTH RATE LOAN PERFORMANCE INFLATION RATE INDEPENDENT VARIABLES DEPENDENT VARIABLE Figure 2.1: Conceptual Framework Table 2.1: Definition of variable under study Variables NPL Definition Measurement NPL (Non-performing loans) are NPL; is a measured ratio of outstanding principal balance of loans past due more loans past due more than 90 days. more than 90 days to outstanding principal balance of all loans. IR IR (Interest rates) is rate at Interest rate; will be measured using which interest is paid by a debtor absolute values obtained from the CBK. for the use of money that they borrow from a lender. RGDP RGDP (Real gross domestic product) RGDP; will be measured using the is an inflation adjusted measure country RGDP values given by the 17 that reflects the value of all goods and CBK and KNBS. services produced in a given year, expressed in base year prices (2009). INFLR INFLR general increase in commodities prices. INFLR; Inflation values from KNBS. 2.4 Empirical Review In recent years, the literature on non-performing loans has occupied the interest of several authors particularly the attention it has generated in understanding of the variables liable to the financial vulnerability (Khemraj and Pasha, 2009). The relationship between the macroeconomic variables and NPLs has also attracted attention in literature (see for example Hoque and Hossain,2004; Siddiqui, Malik and Shah,2011). Among factors cited by the literature as significant determinants include: the Interest rate, the annual Real GDP Growth, the annual Inflation Rate, loans growth, the Unemployment Rate, the Money Supply (M2), the Real Effective Exchange Rate, and Index of Production (Adebola, Yusoff and Dahalan, 2011; Asari et al, 2011; Rinaldi and Sanchis-Arellano, 2006; Salas and Saurina, 2002). In this section, we review the existing literature on the macroeconomic variables affecting loan performance and we will attempt to discuss each variable separately. The variables that were discussed included the studies research objectives: Interest rates, Real GDP Growth Rate and Inflation Rate. Louzis (2012), found that the real GDP growth rate and the interest rates have a strong effect on the level of NPLs, (Louzis et al., 2012, p. 1017). Additionally, the choice of the 18 Inflation Rate relationship with NPLs has been touched on by the work of Warue, (2013) and Fofack, (2005). 2.4.1 Interest rates Interest rates are one of the primary economic determinants of non-performing loans. There is empirical evidence of positive correlation between the interest rate and nonperforming loans (Nkusu 2011; Adebola, Yusoff, & Dahalan, 2011; Louzis, Vouldis and Metaxas, 2011; Berge and Boye, 2007). An increase in interest rate weakens loan payment capacity of the borrower therefore non-performing loans and bad loans are positively correlated with interest rates (Nkusu, 2011). As far as interest rate policy is concerned, it plays a very important role in NPLs growth rate in a country. Hoque and Hossain (2008) examined this issue and according to them, non-performing loans are highly correlated with high interest rates. As a result, this enhances the debt burden of the borrowers and causes loan defaults. Espinoza and Prasad (2010) examined the macroeconomic determinants of non-performing loans in the GCC banking system. According to them, high interest rates increases loan defaults, though they did not find a statistically significant relationship. Bloem and Gorter (2001) studied causes and treatment of NPLs. According to them, frequent changes in the interest rate policy, causes an increase in the NPLs ratio. Asari, et al. (2011) also found significant relationship between loan defaults and interest rates. They also found that an increase in loan defaults also causes asset corrosion of banks and subsequently capital erosion. 19 According to Dash and Kabra (2010) the banks with aggressive lending policies charging high interest rates from the borrowers incur greater non-performing loans. Collins and Wanjau (2011) also found interest rate as a primary factor boosting nonperforming loans. Warue, (2013), investigated the effects of Bank Specific and Macroeconomic Factors on nonperforming Loans in Commercial Banks in Kenya. The study, using panel econometrics approach employing both pooled (unbalanced) panel and fixed effect panel models found that lending interest rates were both positive and significant in affecting non-performing loans in commercial banks. This confirmed previous studies done on the same by Beck, et al., (2013), Souto, et al., (2009), and Aver (2008). This however reflects disparity with Park and Zhang, (2012), who investigated the effects of Macroeconomic and Bank-Specific Determinants of the U.S. NonPerforming Loans: Before and During the Recent Crisis, using two distinct time periods 2002-2006 before the crises and 2007-2010 after the crises and showed that the coefficients for the Federal Funds rate/interest rate was negative in relation to credit risk. 2.4.2 Real GDP Growth Rate Several studies have found GDP growth rate as a significant variable explaining NPLs. The results of studies from most economies of the World, shows a high and negative relationship between NPLs and GDP or real GDP as the case may be. For instance, in a study on the determinants of non-performing loans in the Guyanese banking sector for the period 1994-2004, by Khemraj and Pasha (as cited in Adebola, Yusoff and Dahalan 20 ,2011), the findings reveal among others, evidence of significant inverse and instantaneous relationship between GDP and nonperforming loans, which was interpreted to mean that strong performance in the real economy, results in lower nonperforming loans. Additionally, Beck, et al., (2013), employing estimate fixed-effects and dynamic panel regressions on the basis of annual data for the change in the aggregate NPL ratio, investigated for 75 advanced and emerging economies in the period from 2000 to 2010. The study found GDP rate to have a positive significance effect to non performing loans. This confirms previous studies by Thiagarajan, et al., (2011), Derbali, (2011), Ali and Daly, (2010). The findings are however in sharp contrast with Nkusu (2011), who also analyses the issue with a sample of 26 advanced economies over the period 1998-2009. Using single equation panel regressions and a panel vector autoregressive model, the study found that GDP had a negative relationship on NPLs, and this was further affirmed by the study of Warue, (2013) and Salas and Saurina, (2002). Their studies showed that banks accumulate risks more rapidly in economic boom and some of these risks materialize as asset quality deteriorates during subsequent economic recessions. A study by Simons and Rolwes (2008), found a convincing negative relation between NPLs and GDP growth in the Netherlands. In a study by Dash and Kabra (2010) of Indian commercial banks, using correlation analysis, it was revealed among other findings, that there is a strong negative relationship between NPLs and growth in real GDP. In the same vein, Louzis, Vouldis and Metaxas (as cited 21 in Roland, Petr & Anamaria, 2013), examine the determinants of NPLs in the Greek banking sector and find that credit quality among Greek banks can be explained mainly by macroeconomic fundamentals, among which is GDP. Their economic analysis equally suggests that real GDP growth was the main driver of NPL ratios during the past decade. Therefore, they maintained that a drop in global economic activity remains the most important risk for bank asset quality. However, they were quick to add that economic activity was not able to fully explain the evolution of NPLs across countries and over time. The Espinoza and Prasad (2010) study on the determinants of NPLs in the Gulf Cooperation Council (GCC) banking sector found strong evidence of a significant inverse relationship between real (non-oil) GDP and NPLs. 2.4.3 Inflation Rate Several studies have found inflation rate as a significant variable explaining NPLs. In this regard Mileris (2012) studied the macroeconomic determinants that significantly influence the changes of loan portfolio credit risk in banks and to develop the statistical model for prediction of the proportion of doubtful and non-performing loans. The study employed an OLS regression model for 22 EU countries that were grouped into 3 clusters according to their similarity in changes of the doubtful and non-performing loans percentage in banks for the time period between 2007-2011 and found that an increase in inflation rate had a profound positive relationship to non-performing loans This confirms previous studies by Kochetkov, (2012), Derbali, (2011), Renou, (2011). 22 This was in stark contrast with Warue, (2013), who employed a Comparative Panel Data Analysis using panel econometrics approach employing both pooled (unbalanced) panel and fixed effect panel models, in investigating the effects of Bank Specific and Macroeconomic Factors on nonperforming Loans in Commercial Banks in Kenya. The study came to the findings that inflation was negatively related to credit risk /nonperforming loans. 23 CHAPTER THREE: RESEARCH METHODOLOGY 3.1 Introduction This chapter covered data sources, definition and description of the key variables. It encompassed the research design, target population, sampling frame and sampling technique, data collection procedures, data processing and analysis, and the model specification. 3.2 Research Design Parahoo (1997:142) describes a research design as a plan that describes how, when, and where data are to be collected and analyzed. While Burns and Grove (2003:195) define a research design as a blueprint for conducting a study with total control over factors that may alter the validity of the findings. The research objectives and questions were studied through the use of a descriptive research design. Descriptive research describes data and characteristics about the population or phenomenon being studied. According to Coopers and Schindler (2004) descriptive studies are more formalized and typically structured with clearly stated objectives and research questions. Descriptive research design is going to be employed as it enables the researcher to generalize the findings to a larger population. This study was therefore generalized to all the forty three (43) commercial banks in Kenya. The main focus of this study was quantitative. The choice of the methodology was informed by the data generating process. Previous studies that have used a similar research design 24 include: Gremi, (2013), Park and Zhang, (2012), Mileris, (2012), Castro, (2012), Renou, (2011), Igan andPinheiro, (2011), Vogiazas & Nikolaidou, (2011), Salas and Saurina, (2002). 3.3 Target Population and Sampling. A target population is any set of persons or objects that possesses at least one common characteristic (Busha & Harter, 1980). There are 43 commercial banks in Kenya Central Bank of Kenya, (2015). This study used time series data from all the 43 commercial banks in Kenya to avoid the sampling bias problem covering the monthly period from 2008 to 2015 third quarter. All these banks were studied since a conclusive and whole representative analysis was to be arrived at in the end. 3.4 Data Collection Procedures This study used annual data from the Central Bank of Kenya, World Bank and Kenya National Bureau of Statistics (KNBS) covering the period from 2008 to 2015 third quarter. The study employed the use of secondary data in its analysis from the aforementioned sources. Interest rates were also used in the study as an independent variable for the period from 2008 to 2015 third quarter. It is defined as the amount charged, expressed as a percentage of principal, by a lender to a borrower for the use of assets. Monthly data was used to apply consistence with the other data formats for other variables employed in the study. 25 The Real GDP Growth Rate variable in this study was one of the four independent variables for the equation which showed the relationship between it and non-performing loans, the dependent variable. Monthly data from 2008 to 2015 third quarter was used, and the data was sourced from the World Bank website and from the Central Bank of Kenya. Inflation is an increase in the general price level and is typically expressed as an annual percentage rate of change. Inflation is important for banks in their capacity of financial intermediation having adjusted for anticipated inflation, and can suffer massive default risk depending on the fluctuation of inflation between the anticipated and actual inflation rates on their fixed instruments Glogowski, (2008). Rising inflation tends to lead to an increase in non-performing loans. Monthly inflation rate data for this study was sourced from the Kenya National Bureau of Statistics (KNBS) from 2008 to 2015 third quarter. 3.5 Data Processing and Analysis According to Babbie (2010), data analysis is carried on the data collected to transform it to a form that is suitable for use in drawing conclusions that reflect on the ideas, and theories that initiated the inquiry. In order to explain the macroeconomic determinants of NPLs in commercial banks in Kenya, I used the ordinary least squares model (OLS). Under this approach it needs to be considered that the OLS’s main assumption is that the errors must be uncorrelated. Considering that the model involves time series, this assumption could be violated since it is reasonable to think that the cyclical effect in the 26 economy indicates positive autocorrelation. However, previous studies have shown that OLS is a suitable model to describe NPL time series (For instance, Espinoza R. and Prasad A. (2010) and Bofondi, M. and Ropele T, (2011)) as long as the existence of autocorrelation in residuals is investigated. Other OLS model assumptions include, linearity of the model, its non-stochastic characteristic, having mean value of 0, and distribution with equal variance etc. 3.6 Model Specification (Time series model) Consistent with, Brooks, (2008), the econometric model is specified; NPLt = α+ β1Δ IRt + β2Δ RGDPt + β3Δ INFLRt + εt Where, NPLt =Non-Performing Loans at time, t IRt =Real Interest Rate at time, t RGDPt =Real GDP Growth Rate at time, t INFLRt =Inflation Rate at time, t εt = the error term is assumed to be normally and independently distributed with mean zero and constant variance, at time, t αt = the intercept coefficient estimate, is interpreted as the value that would be taken by the dependent variable y if the independent variable x took a value of zero. 27 β1, β2, β3, are the coefficients of the various explanatory variables, measured with respect to IR, RGDP, and INFLR respectively. Previous studies have shown that OLS is a suitable model to describe NPL time series Espinoza & Prasad, (2010). 28 CHAPTER FOUR: RESEARCH RESULTS AND DISCUSSIONS 4.1 Introduction This chapter presents the results and findings of the study, which was to establish the macroeconomic factors affecting loan performance of commercial banks in Kenya. The analysis is based on monthly data collected from 2008 to 2015 third quarter. The results are presented in the form of summary tables. The data for this study was obtained from Central Bank of Kenya and Kenya National Bureau of Statistics. The data was tested for normality and stationality of the variables, correlation analysis and co-integration regression of the variables. Post-estimation tests were also carried out for normality of the residuals, autocorrelation and Heteroskedasticity. 4.2 Univariate Analysis Time series data often have time-dependent moments (e.g. mean, variance, skewness, and kurtosis). When analyzing time series data, the initial step was to investigate whether the variables under study are normally distributed. To test for normality of the variables, descriptive statistics was undertaken putting keen interest on the Jarque-Bera (JB) probability. Our concern being also on the measures of central tendency that comprises of the mean, median, and the mode as well as the measures of variability or dispersion that comprises of standard deviation (or variance). When using the jargueBera (JB) test, a null hypothesis of normal distribution was tested against the alternative 29 hypothesis of non-normal distribution. For normal distribution the JB statistics is expected to be statistically indifferent from zero thus; Ho: JB=0 (Normally distributed) H1: JB≠0 (Not normally distributed) Rejection of the null for any of the variables would imply that the variables are not normally distributed. The result (Table 4.1) shows that a total of 90 occurrences of each variable were used in the study. The result indicates that the overall average ratio of gross non-performing loans for commercial banks in Kenya under the study was 6.31%, interest rate was 14.76%, the real GDP growth rate was 4.15% and the inflation rate was 9.83%. Table 4.1 shows that all the variables were not normally distributed. The findings therefore show features of non-normality which is common in financial time series data. As such, Table 4.1 presents the results of descriptive statistics of all the variables for the period ranging from 2008 to third quarter 2015. Table 4.1 Descriptive Statistics NPLS IR RGDP INFLR Mean 0.0631 0.1476 0.0415 0.0983 Median 0.0554 0.1402 0.0440 0.0699 Maximum 0.0996 0.1848 0.0592 0.2739 Minimum 0.0426 0.1284 0.01 0.0313 Std. Dev. 0.0169 0.0171 0.0134 0.0663 30 Skewness 0.6189 0.8551 -0.7586 1.1570 Kurtosis 1.9869 2.6095 2.5961 3.2165 Jaque-Bera 9.5933 11.5409 9.2438 20.2558 Probability 0.0083 0.0031 0.0098 0.00004 Sum 5.6777 13.2853 3.7361 8.8530 Sum Sq. Dev. 0.0253 0.0259 0.0159 0.3909 Observations 90 90 90 90 4.3 Correlation Analysis The correlation analysis is statistical technique employed to measure the strength or degree of linear association between two variables. Correlation analysis is used to check for multicollinearity between the variables. Multicollinearity is a serious problem if the correlation coefficient between two regressors is above 0.8. The correlation coefficient can range from -1 to +1 with -1 indicating a perfect negative correlation +1 indicating a perfect positive correlation , and 0 indicating no correlation at all. Table 4.2 shows the relationship between the dependent and independent variables. From the correlation matrix, NPLS are negatively correlated with IR (-0.684 or 68.4%). NPLS are also negatively correlated with RGDP (-0.597 or 59.7%). Additionally, NPLS are positively correlated with INFLR (0.419 or 41.9%). The table also shows the relationships between the independent variables. Interest rates had a mild positive correlation with RGDP (0.264 or 26.4%), and a mild negative correlation with INFLR (-0.279 or 27.9%). Real GDP had relatively strong negative correlation with inflation rate (-605 or 60.5%). The correlation table shows that all the variables can be included in the same model since 31 they are below the multicollinearity level of 0.8. These low correlations also give the signal of no multicollinearity among the variables. Table 4.2 Correlation NPLS IR RGDP INFLR NPLS 1.000000 -0.683931 -0.596578 0.419327 IR -0.683931 1.000000 0.263594 -0.279234 RGDP -0.596578 0.263594 1.000000 -0.604520 INFLR 0.419327 -0.279234 -0.604520 1.000000 4.4 Stationarity Test When estimating a model of time series variables, it is necessary to ensure that all time series variables are stationary. The mean or variance of many time series increases over time. This is a property of time series data called non stationarity. As Granger and Newbold,(1974) demonstrated, if two independent, non-stationary series are regressed on each other, the chances for finding a spurious relationship are very high. Shocks (e.g. the 2007/08 ethnic clashes) to a stationary series are temporary; the series reverts to its long run mean. For non-stationary series, shocks result in permanent moves away from the long run mean of the series. Augmented Dickey-Fuller (ADF) and the Phillips Perron (PP) tests were conducted and the results reported in Table 4.3 and 4.4 respectively. The decision criterion involves comparing the computed ADF and PP statistic values with the critical values at 5%. If the computed ADF and PP statistic is 32 greater in absolute terms compared to the critical values, the null hypothesis of nonStationarity in time series variables is rejected and vice versa. Table 4.3: Stationality test (Augmented Dickey-Fuller) Variable 1st Difference Levels Constant Trend and Constant Integration Trend and Intercept Intercept t-stat 5% t-stat 5% t-stat 5% t-stat 5% NPLS -2.29 -2.89 -0.737 -3.46 -10.55 -2.89 -11.25 -3.46 1(1) IR -1.73 -2.89 -1.61 -3.46 -7.11 -2.89 -7.12 -3.46 1(1) RGDP -2.39 -2.89 -2.89 -3.46 -8.76 -2.89 -8.73 -3.46 1(1) INFLR -1.84 -2.89 -1.88 -3.46 -6.14 -2.89 -6.13 -3.46 1(1) Table 4.4: Stationality test (Phillips Perron) Variable 1st Difference Levels Constant Trend and Constant Integration Trend and intercept intercept t-stat 5% t-stat 5% NPLS -2.33 -2.89 -0.587 IR -1.77 -2.89 -1.69 RGDP -2.88 -2.89 -3.54 -3.46 INFLR -1.78 -2.89 -1.92 -3.46 -3.46 -3.46 t-stat 5% t-stat 5% -10.49 -2.89 -11.19 -3.46 1(1) -7.26 -2.89 -7.26 -3.46 1(1) - - - -2.89 -6.15 -6.27 33 1(0) -3.46 1(1) When variables are integrated at different levels as above, the first task is to test for cointegration of the variables that are integrating at 1. The results are reported in Table 4.5. It is well known that if two series are integrated at two different orders, linear combination of them will be integrated to the higher order of the two orders. But it is possible that certain combinations of the non-stationary series are stationary. Then it is said that the pair yt, xt are co-integrated. The results show that the variables are integrating and therefore we can run a regression. Table 4.5: Johansen Cointegration test Unrestricted Cointegration Rank Test (Trace) Hypothesized Trace 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.** None * 0.315083 59.60430 47.85613 0.0027 At most 1 * 0.134221 26.29997 29.79707 0.1200 At most 2 * 0.091478 13.61695 15.49471 0.0941 At most 3 * 0.057107 5.174609 3.841466 0.0229 Trace test indicates 1 cointegrating eqn(s) at the 0.05 level * denotes rejection of the hypothesis at the 0.05 level **MacKinnon-Haug-Michelis (1999) p-values 34 Unrestricted Cointegration Rank Test (Maximum Eigenvalue) Hypothesized Max-Eigen 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.** None * 0.315083 33.30433 27.58434 0.0082 At most 1 * 0.134221 12.68303 21.13162 0.4818 At most 2 0.091478 8.442339 14.26460 0.3355 At most 3 * 0.057107 5.174609 3.841466 0.0229 Max-eigenvalue test indicates 1 cointegrating eqn(s) at the 0.05 level * denotes rejection of the hypothesis at the 0.05 level 1 cointegrating Equation(s): Log likelihood 1351.804 Normalized cointegrating coefficients (standard error in parentheses) LNPLS 1.000000 LIR LRGDP LINFLR 0.852381 1.276662 0.055787 (0.12179) (0.18835) (0.03819) 4.5: Cointegration Analysis The co-integration results as shown in table 4.5 using the Johansen co-integration test show that there was 1 co-integrating equation according to the trace test and the Eigen test. Therefore, there was a long run relationship and equilibrium on the variables. The 35 existence of co-integrating vectors imposes the transformation of the OLS model into a VECM model to analyze the dynamic inter-relationships as shown on table 4.6. Table 4.6: Vector error correction model Variable Coefficient LNPLS 0.047290 0.024786 1.907923 0.0599 LNPLS -0.215168 0.113128 -1.901989 0.0607 LIR 0.027541 0.068675 0.401035 LRGDP -0.087671 0.040961 -2.140350 0.0353 LINFLR 0.001542 0.021002 0.073410 0.9417 C (6) -0.000501 Std Error 0.000285 t-statistics -1.757410 Prob 0.6894 0.0826 R-Squared 0.090330 F-Statistics 1.628526 Adjusted R-Squared 0.034863 Prob (F-Statistic) 0.161654 S.E of regression 0.002614 Akaike Info Criterion -8.990047 Sum Squared residual 0.000560 Durbin-Watson Stat 1.90 4.6 Vector error correction model analysis The vector error correction results from table 4.6 shows that the only significant explanatory variable was LRGDP (-1) with a probability value of 0.0353 or 3.53%. The coefficient value of LRGDP (-1) was a negative number (-0.087671). This showed that in a 1 lag monthly period, LRGDP was negatively correlated to NPLS. A plausible interpretation of these results was that an increase in the real GDP growth rate in the short term (1 lag monthly period), can be explained by financial theory. The theory 36 suggests that, in prosperous times, when corporate cash flow rises beyond what is needed to pay off debt, a speculative euphoria develops, and soon thereafter debts exceed what borrowers can pay off from their incoming revenues, which in turn produces a financial crisis. As a result of such speculative borrowing bubbles, banks and lenders tighten credit availability, even to companies that can afford loans and the economy subsequently contracts, leading to the resultant NPLS. The results of studies from most economies of the World, shows a high and negative relationship between NPLs and GDP or real GDP as the case may be. For instance, in a study on the determinants of non-performing loans in the Guyanese banking sector for the period 1994-2004, by Khemraj and Pasha (as cited in Adebola, Yusoff and Dahalan ,2011), the findings reveal among others, evidence of significant inverse and instantaneous relationship between GDP and nonperforming loans, which was interpreted to mean that strong performance in the real economy, results in lower non-performing loans. The table also shows that the other explanatory variables IR and INFLR, were not significant in explaining NPLS. Further table 4.6 shows that there is no evidence of autocorrelation since the Durbin-Watson d statistics which tests for 1st order correlation was approximately 2 (1.90). 37 Table 4.7: Summary of residual diagnostic tests Normality Test Jarque-Bera statistic 5.898522 Probability 0.0524 Breusch-Godfrey Serial Correlation LM Test F-Statistic * 1.452205 Prob. F (2,80) Obs R-squared 3.082925 Durbin-Watson Stat 1.965 0.2402 Prob. Chi-Square (2) 0.2141 Heteroskedasticity Test: White Test F-Statistic 1.048965 Prob. F (20,67) 0.4217 Obs*R-squared 20.98424 Prob. Chi-Square (20) 0.3981 Scaled explained SS 27.20200 Prob. Chi-Square (20) 0.1297 4.7 Discussion of residual diagnostic tests Table 4.7 shows the summary of residual diagnostic tests. The first upper portion tested on Jarque-Bera normality test and showed that the residuals were normal with a probability (P-Value) of 0.0524 (5.24%), greater than 0.5%. The second upper portion tested on the Breusch-Godfrey serial correlation LM test. In this portion, the Chi-Square probability value was 0.21, greater than 0.5. As such, the residuals were not serial correlated. On the other hand, the Durbin-Watson was used to test for the 1st order correlation and found no correlation since the DW was approximately 2 (1.965). 38 The third upper portion heteroskedasticity tests using the White test revealed the absence of heteroskedasticity in the residuals, and as a result concluded that the residuals were homoskedastic in nature. Table 4.8: OLS model equation estimate Variable Coefficient C Std Error t-stat Prob 0.170132 0.011081 15.35387 0.0000 LIR -0.561265 0.065777 -8.532826 0.0000 LRGDP LINFLR -0.573955 -0.003770 0.101017 0.020510 -5.681755 -0.183816 0.0000 0.8546 R-Squared 0.6441 Mean dependent var 0.063085 Adjusted R-Squared 0.642 S.D dependent var 0.016867 S.E of regression 0.010091 Akaike Info criterion -6.310889 Sum Sq. residual 0.008757 Schwarz criterion -6.199787 Hannan-Quinn criterion -6.266086 Durbin-Watson Stat 0.308751 Log likelihood F-Statistic Prob (F-Stat) 287.9900 54.21895 0.000000 4.8: Results Discussion Our results from the basic OLS model equation estimate, suggest that LIR and LRGDP had a negative and significant relationship with the gross ratio of LNPLs. This is an overwhelming support for our research objectives. A plausible interpretation of these results is that an increase in interest rates (LIR) resulted in banks increasing their 39 provisions of loan loss that inevitably decreased the banks revenue and subsequently reduced the funds for new lending with the resultant reduction in LNPLS (Hou, Y, 2007). The findings also showed that while IR (Un-logged) during the first few months of 2012 were at their highest at 20%, while nonperforming loans (Un-logged NPLS) reported the lowest figure of 4.33% as shown in appendix 11 in the shaded area. According to Muniu (2013) study on the relationship between changes on lending rates and the level of nonperforming loans in commercial banks in Kenya, he concluded that bank rates have a negative significant effect on the nonperforming loans by way of a decrease in gross ratio of NPLs of banks since there will be improved economic conditions of both households and corporate, hence they will be able to repay their borrowed amounts owing to the improved economy. This is also the case for LRGDP in our OLS model estimate. From our model the estimates show that inflation rate is not significant with NPLs. This was also inconformity with Warue, (2013). This implies that Non-performing loans are not responsive to changes in inflation. A probably interpretation of these results is that inflation leads to more profitability as more money chases few goods. Most borrowers are business people who seem to pass over the cost of inflation to consumers. For instance, when fuel prices go up, road transport players raise fare to consumers of their services. Thus business people retain their ability to repay their loans. The OLS model also showed an R-Squared coefficient of 0.6441, implying that the model was a good fit and that two out of the three explanatory variables were 40 significant. The Prob (F-Stat) also reviewed that the variables were significant and a good fit for the model. 41 CHAPTER FIVE: SUMMARY, CONCLUSIONS AND RECOMMENDATIONS 5.1 Introduction This chapter presents a summary of the major findings that are presented in chapter four, summary conclusions and recommendations suggested by the researcher. All findings are summarized in line with the objectives and shows how the objectives have been achieved. 5.1 Summary The study sought to establish the effects of real GDP growth rate, interest rates, and inflation rate on the gross NPLs ratio of commercial banks in Kenya using the OLS model estimate and the correlation matrix, which showed the relationships between the dependent variable and the explanatory variables. The study used all the 43 commercial banks in Kenya and found that interest rate and real GDP growth rates had a significant and negative relationship with NPLs gross ratio, and as such were found to be important and significant in explaining NPLs in commercial banks in Kenya. Inflation rate was found to be insignificant in explaining NPLs in commercial banks in Kenya. 5.3 Conclusions The first objective of the study was to establish the effect of interest rates on gross NPLs ratio in commercial banks in Kenya. The findings indicated that interest rates had a negative and significant effect on gross NPLs ratio in commercial banks in Kenya. This 42 implied that an increase in the lending rate on commercial banks caused banks to increase their provisions of loan loss that invariably decreased the banks revenue and reduced the funds for new lending and subsequently, reduced the gross ratio of NPLs. The second objective of the study was to establish the effect of real GDP growth rate on gross NPLs ratio in commercial banks in Kenya. The findings indicated that real GDP growth rate had a negative and significant effect on the gross NPLs ratio. This implied that continued improvement of the economy would see households and corporate easily repay their loans due to improved economic conditions. The third objective of the study was to establish the effect of Inflation on gross NPLs ratio in Kenyan commercial banks. The findings indicated that Inflation was insignificant in explaining the NPLs gross ratio in Kenyan commercial banks. 5.4 Recommendations The study sought to provide more information to the Central Bank, Managers of commercial banks and Investors on the effects of varying macroeconomic factors on gross NPLs ratio so as to make informed decisions. Following the conclusions drawn from the above findings of the study, it is recommended that in that the Central Bank of Kenya should play a more regulatory and active role by ensuring that banks have adequate provisions for bad loans, that they also play an active role in forecasting of macroeconomic changes hence can alert banks well enough of time so that they can input the changes/ shifts into their operations. Banks should keep the interest rate as low 43 as possible to make sure that repayment is easy and affordable for the holders of loans, the banks should consider fixed lending interest rates for huge sums of money so as to minimize the default amount. Another recommendation would be for banks to use a reducing balance approach on loans this would greatly reduce the rate of loan defaulters and ultimately reduce credit risk. The banks should shift their mindset of earning much of their income from interest loans and divest into bank charges and voluminous loan holders through cheaper lending rates while still applying stringent security checks for loan applicants and holders and emphasis should be put on the credit history of people up taking loans so as to safeguard against credit default. Finally, interest rate and real GDP growth rate are essential and critical components in forecasting and explaining NPLs growth. 5.5 Suggestions for Further Research The effect of Macroeconomic variables on gross NPLs ratio in Kenyan commercial banks is a research area where a lot of research has not been carried out and also varying results have been published on the same. Further, non-testable endogenous variables like asymmetric information and credit criteria as relevant variables explaining NPLs in Kenya have not been extensively explored. 44 References Adebola, S.S., Wan Yusoff, S.B., & Dahalan, D.J. (2011). An approach to the determinants of non-performing loans. Kuwait Chapter of Arabian Journal of Business and Management Review, 1 Ahmed, J.U. (2010). An empirical estimation of loan recovery and asset quality of commercial banks, The NEHU Journal, 8 Akerlof, G.A. (1970). The market for “lemons”: Quality uncertainty and the market mechanism, The Quarterly Journal of Economics, 84(3) Alary, D & Goller, C. (2001). Strategic default and penalties on the credit market with potential judgement errors, EUI working paper Asari, F.F., Muhammad, N.A., Ahmad, W., Latif, N.I., Abdulllah, N. & Jusoff, K. (2011). An analysis of non-performing loans, interest rates and inflation rate, World Applied Sciences Journal Barr, R.S. & Siems, T.F. (1994). Predicting bank failure using DEA to quantify management quality, Fiancial Industry Studies, (1), 1-31 45 Barron, J.M. & Staten, M.E. (2008). The emergence of captive finance companies and risk segmentation in loan markets: Theory and evidence, Journal of Money, Credit and banking, 40, 173-192 Bester, H. (1994). Collateral, default risk and relationship lending: An empirical study on financial contracting, Journal of Money, Credit and Banking, 26(1), 72-86 Bexley & Nenninger. (2012). Financial institutions and the economy, Journal of Accounting and Finance, 12 Bloem, A.M. & Gorter, C.N. (2001). The treatment of non-performing loans in macroeconomic statistics, IMF Working Paper, 1, 209 Bofandi, M. & Gobbi, G. (2003). Bad loans and entry in local credit markets, Bank of Italy Research Department, Rome Brownbridge, M. (1998). The cause of financial distress in local banks in Africa and implications for prudential policy, UNCTAD/OSG/DP/132 Brown, M. & Zehnder, C. (2008). The emergence of information sharing in credit markets, Swiss National Bank Working Papers Chand, S. (2002). Financial sector development and economic growth in pacific countries, Pacific Economic Bulletin, 17(1), 117-133 46 Choudhury et.al (2005). Non-performing loans in the banking sector of Bangladesh: Realities and challenges, Bangladesh Institute of Bank Management Collins, N.J. & Wanjau, K. (2011). The effects of interest rate spread on the levels of non-performing assets: A case of commercial banks in Kenya, International Journal of business and Public Management, 1(1) Crowley, J. (2007). Interest rate spread in English speaking African countries, IMF Working Paper, 7,101 Dash, M. & Kabra, G. (2010). The determinants of non-performing assets in Indian commercial banks: An econometric study, Middle Eastern Finance and Economics Journal, 7, 94 Ekumah, E.K. & Essel, T.T. (2003). Information is power: The problem with credit accessibility in rural banks in Ghana, International Labor Organization, 2002 Espinoza, R. & Prasad, A. (2010). Non-performing loans in the GCC banking system and their macroeconomic effects, IMF Working Paper, 10, 224 47 Fofack & Hippolyte. (2005). Non-performing loans in Sub-Saharan Africa: Causal analysis and macroeconomic implications, World Bank Policy Research Working Paper Greenidge, K. & Grosvenor, T. (2010). Forecasting non-performing loans in Barbados, Journal of Business, Finance and Economics in Emerging Economies, 5, 80107 Hoque,M.Z. & Hossain, M.Z. (2008). Flawed interest rate policy and loan default: Experience from developing country, International Review of Business Research Papers, 5(5), 235-246 Kalberg, J. & Udell, G. (2003). Private sector credit information: The U.S case, Credit Reporting Systems and the International Economy Edition Kithinji, A. & Waweru, N.M. (2007). Merger restructuring and financial performance of commercial banks in Kenya, Economic, Management and Financial Markets Journal, 2(4), 9-39 Klein, D.B. (1992). Promise keeping in the great society: A model of credit information sharing, Economies and Politics, 4(2), 117-136 Kroszner, P. (2002). Non-performing loans, monetary policy and deflation: The industrial country experience, Economic and Social Research Institute, Japan 48 Leland, H.E. & Pyle, D.H. (1977). Informational asymmetries, financial structure and financial intermediation, Journal of Finance, 32, 371-387 Louzis, D.P., Vouldis, A.T. & Metaxas, V.L. (2011). Macroeconomic and bank-specific determinants of non-performing loans in Greece: A comparative study of mortgages, business and consumer loan portfolios, Journal of Banking and Finance Marco, P. & Jappelli, T. (1993). Information sharing in credit markets, The Journal of Finance, 43(5), 1693-1718 McNulty, J., Akhigbe, A. & Verbrugge, J. (2001). Small bank loan quality in a deregulated environment: The information advantage hypothesis, Journal of Economics and Business,53, 325-339 Michael, R. & Joseph, S. (1976). Equilibrium in competitive insurance markets: An essay on the economics of imperfect information, The Quarterly Journal of Economics, 90(4), 629-649 Ndung`u, N.S. & Ngugi, R.W. (2000). Banking sector interest rate spread in Kenya, Kippra Discussion Paper, 5 Ngugi, R.W. (2001). An empirical analysis of interest rate spread in Kenya, AERC Research Paper, 106, 2-34 49 Nkusu, M. (2011). Non-performing loans and macro-financial vulnerabilities in advanced economies, IMF Working Paper, 11, 161 Padilla, A.J. & Marco, P. (1996). Sharing default information as a borrower discipline device, Industry Study Program Discussion Paper, 73 Padilla, A.J. & Marco, P. (1997). Endogenous communication among lenders and entrepreneurial incentives, The Review of Financial Studies, 10(1), 205236 Pauly, M.V. (1968). The economics of moral hazard, American Economic Review, 58, 531-537 Waweru, N.M. & Kalani, V.M. (2009). Banking crisis in Kenya: Causes and remedies, African Journal of Finance and Banking Researc, 4(4), 12-23 50 APPENDIX 1: STATIONALITY GRAPHS NPLS IR 1.11 1.22 1.10 1.20 1.09 1.08 1.18 1.07 1.16 1.06 1.14 1.05 1.04 1.12 08 09 10 11 12 13 14 15 08 09 10 RGDP 11 12 13 14 15 13 14 15 INFLR 1.07 1.35 1.06 1.30 1.05 1.25 1.04 1.20 1.03 1.15 1.02 1.10 1.01 1.05 1.00 1.00 08 09 10 11 12 13 14 15 08 51 09 10 11 12 APPENDIX II Sample Data Collected And Transformed (1+r) IR 1.138 1.138 1.141 1.139 1.14 1.141 1.139 1.137 1.137 1.141 1.143 1.149 1.148 1.147 1.149 1.147 1.149 1.151 1.148 1.148 1.147 1.148 1.148 1.149 1.148 1.148 1.15 1.148 1.146 1.145 1.144 1.143 1.142 1.14 1.139 RGDP 1.011 1.01 1.012 1.02 1.021 1.022 1.027 1.026 1.027 1.015 1.014 1.015 1.06 1.06 1.061 1.02 1.021 1.02 1.021 1.023 1.022 1.027 1.028 1.029 1.047 1.047 1.048 1.046 1.047 1.046 1.061 1.06 1.059 1.059 1.058 INFLR 1.182 1.191 1.218 1.226 1.315 1.293 1.265 1.276 1.282 1.289 1.294 1.277 1.219 1.146 1.146 1.124 1.096 1.086 1.084 1.073 1.067 1.066 1.05 1.053 1.047 1.052 1.045 1.037 1.039 1.035 1.036 1.0322 1.0321 1.0318 1.0384 52 NPLs 1.1038 1.1029 1.1047 1.0946 1.0955 1.0975 1.0901 1.0885 1.0878 1.0845 1.0844 1.09 1.089 1.0884 1.0935 1.0931 1.0988 1.0901 1.0899 1.0894 1.0823 1.0816 1.0812 1.0794 1.0819 1.0809 1.0798 1.0775 1.0763 1.079 1.0717 1.0704 1.0696 1.0681 1.065 1.14 1.139 1.14 1.139 1.139 1.139 1.139 1.139 1.141 1.143 1.148 1.152 1.185 1.2 1.195 1.203 1.203 1.202 1.201 1.203 1.202 1.201 1.197 1.19 1.178 1.182 1.181 1.178 1.177 1.179 1.175 1.17 1.17 1.17 1.169 1.17 1.169 1.17 1.17 1.057 1.049 1.051 1.05 1.04 1.041 1.04 1.04 1.039 1.04 1.044 1.043 1.044 1.04 1.041 1.04 1.044 1.045 1.045 1.045 1.045 1.046 1.046 1.046 1.046 1.051 1.05 1.053 1.047 1.046 1.045 1.045 1.045 1.046 1.039 1.039 1.039 1.044 1.045 1.0451 1.0542 1.0654 1.0919 1.1205 1.1295 1.1449 1.1553 1.1667 1.1732 1.1891 1.1972 1.1893 1.1831 1.1669 1.1561 1.1306 1.1222 1.1005 1.0774 1.0609 1.0532 1.0414 1.0325 1.032 1.037 1.045 1.041 1.0405 1.0411 1.049 1.06 1.067 1.083 1.078 1.074 1.071 1.072 1.069 53 1.0624 1.0575 1.0609 1.0601 1.0591 1.057 1.0538 1.0518 1.0505 1.0484 1.05 1.0459 1.044 1.0491 1.0446 1.0435(2012 1.0443 1.0438 1.0446 1.0449 1.0452 1.0455 1.0457 1.046 1.0463 1.046 1.047 1.05 1.053 1.056 1.053 1.053 1.053 1.052 1.053 1.051 1.05 1.055 1.058 1.171 1.169 1.167 1.17 1.164 1.169 1.163 1.16 1.16 1.159 1.16 1.159 1.155 1.155 1.154 1.153 1.155 1.158 1.158 1.044 1.057 1.058 1.058 1.055 1.054 1.056 1.055 1.056 1.055 1.06 1.06 1.056 1.056 1.056 1.056 1.063 1.064 1.073 1.074 1.077 1.084 1.066 1.064 1.061 1.06 1.0553 1.056 1.063 1.07 1.071 1.073 54 1.056 1.057 1.056 1.057 1.055 1.054 1.054 1.054 1.054 1.054 1.056 1.057 1.057 1.057 1.059 1.061