- Email: [email protected]
Efficiency of financial market intermediation in Kenya: A comparative analysis
Available online at www.sciencedirect.com
Journal of Policy Modeling 33 (2011) 226–240
Efficiency of financial market intermediation in Kenya: A comparative analysis Jacob Oduor a,∗ , Stephen Karingi b , Stephen Mwaura c a
Kenyatta University and the Kenya Institute for Public Policy Research and Analysis, Kenya, P.O. Box 56445-00200 Nairobi, Kenya b United Nations Economic Commission for Africa, Addis-Ababa, Ethiopia c Central Bank of Kenya, Kenya Received 1 March 2010; received in revised form 1 July 2010; accepted 1 August 2010 Available online 21 September 2010
Abstract Wide interest margins as witnessed in Kenya are a sign of a repressed and inefficient financial sector. This paper carries out a cross-country analysis of the determinants of financial market efficiency using panel cointegration with a view to recommending policy options for improving the efficiency of the financial sector intermediation process in Kenya. The study finds that the major contributors to the differences in financial sector inefficiency in Kenya compared to the other countries in the study are high bank operating costs, default risk and financial market structure. The study recommends, among other measures, that the government through the Central Bank need to collaborate with the commercial banks and establish a working credit reference bureau to enable easy identification of credit worthy customers in order to reduce default risk; there is also need by the central bank to license more new banks to increase competition and reduce bank concentration. The study also recommends increased use of technology including phone-banking and e-banking to reduce operation costs of the banks. The paper concludes that contrary to the findings from other cross-country analysis, the factors that lead to financial market in/efficiency varies from one country to the other. © 2010 Society for Policy Modeling. Published by Elsevier Inc. All rights reserved. JEL classification: G14; G21; C33 Keywords: Financial market efficiency; Panel cointegration
∗
Corresponding author. Tel.: +254 722978186. E-mail addresses: [email protected], [email protected] (J. Oduor).
0161-8938/$ – see front matter © 2010 Society for Policy Modeling. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.jpolmod.2010.09.003
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
227
1. Background The ability of the financial market to efficiently allocate and reallocate loanable funds is critical for economic growth. When financial markets are efficient, the resultant competition among the financial institutions and reduced transaction and operation costs will imply higher deposit rates and lower lending rates as the financial institutions compete to attract borrowers and savers. If competition for instance is lacking in the market, there is a possibility that lending rates may be higher than the free market average and deposit rates will be lower than the free market rates. Would-be depositors will therefore not save as much. They may instead opt to consume and not save. This stifles savings mobilization and reduces funds available for investments and leads to misallocation of resources in the economy. On the other hand, investors will shy away from borrowing as the lending rates rise. Since the main role of financial institutions is to intermediate between the borrowers and lenders in the market, high lending rates and low deposit rates, and by implication, wide interest rate spreads, are indications of inefficient financial sector intermediation process. Interest rate spread is the difference between lending rates and deposit rates and has been used severally in literature to indicate the efficiency of the financial sector intermediation process (see for instance Demirguc-Kunt & Hiuzinga, 1997). Widening spreads therefore indicate increasingly inefficient financial market. Discussions on interest rate spreads in Kenya have dominated many forums on economics and business in the recent past. The point is clear—lending interest rates in Kenya though have been falling from 2003 still remains very high while deposit rates are very low. This, as can be seen in Fig. 1 in Appendix B, has left interest rate margins relatively high and unchanged. At the gist of this discussion is the concern about the consequences that such high interest rates spreads have on economic performance of the country and on the welfare of the households and firms who must borrow to invest or save to consume in future. This concern puts to question the ability of the financial sector in Kenya to mediate efficiently between borrowers and lenders in the economy. The wide margins as witnessed in Kenya, is a sign of financial sector repression, which compromises the financial sector’s role in the intermediation process of deposit mobilization and credit creation. The other concern is the implication of such wide margins on Kenya’s global competitiveness. The commercial banks in Kenya levy over 12 percentage points spread between the deposit and lending rates as opposed for instance to 3.7% in South Africa, 3% in Singapore and 1% in Korea (Banking Survey, 2007). These differences jeopardize Kenya’s international competitiveness as international competitors are at an advantage of at least 4% points in their cost of production against their Kenyan counterparts. A comparison of some of the fundamental determinants of interest rates with those of other countries shows no significant differences. For instance, banking sector concentrations in South Africa and Egypt are almost similar to Kenya’s yet interest margins in the two countries are much lower than in Kenya. The questions that come to mind therefore are ‘why are interest margins so wide in Kenya compared to the other countries and what are the driving factors?’ In light of these concerns, it is important therefore to examine the relative importance of the determinants of interest rate margins in Kenya vis a vis those of other countries to determine the factors that explain the high margins and are specific to Kenya and not the other countries. Studies on interest rate margins in Kenya like Ndung’u and Ngugi (2000) have not analyzed cross-country differences in the financial sector determinants of interest rates spreads to establish whether the determinants significantly differ across countries. In addition, Ndung’u and Ngugi (2000) did not capture the influence of non-performing loans (credit risk), market power or structure, and the transaction costs which are very important variables in explaining the financial market efficiency. This study, in addition to filling this knowledge gap, attempts to inform policy
228
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
on the major focus areas in line with the significance of the fundamentals in the other countries as compared to their significance in Kenya. To analyze the main determinants of interest rate spreads as an indicator of financial sector efficiency, we use panel cointegration method with data from 1990 to 2007 for 11 countries. The results show that the major contributors to the financial sector inefficiency in Kenya compared to the other countries in the study are high operating costs (cost inefficiency), default risk or level of non-performing loans and bank size or bank concentration. In terms of magnitude, cost inefficiency is the major contributor to financial market inefficiency in Kenya followed by financial market structure, and default risk in that order. Further the findings show that the factors that lead to market inefficiency in one country are not necessarily the same as the factors that lead to market inefficiency in the other countries. The study recommends, among other measures, that the government through the Central Bank and the commercial banks need to collaborate and establish a working credit reference bureau to identify easily credit worthy customers in order to reduce default risk and also stresses on the need by the central bank to license more new banks to increase competition and reduce bank concentration. There is also need to stress on increased use of technology including phone-banking and e-banking to reduce operation costs of the banks. The rest of this paper is organized as follows; the remainder of Section 1 gives the comparisons of interest rate spreads and its determinants in Kenya with those of other countries. Section 2 gives the literature review, Section 3 gives the empirical framework adopted and the diagnostic tests while Section 4 discusses the empirical results. Section 5 concludes and gives policy recommendations suggested. 1.1. Comparisons of some of the fundamental determinants of interest margins Comparisons of Kenyan margins with those of some of Kenya’s strongest competitors in the export markets reveals great disparities. The comparisons show quite lower margins for the other countries including among others; Korea, and Singapore (see Fig. 2 in Appendix B) and even regionally including South Africa and Egypt (see Fig. 3 in Appendix B). Whereas Kenya had average interest rate margins of 11.5% in 2007, Kenya’s main competitors South Africa have 3.7% and Egypt on the other hand has interest margins of 3.2%. This puts these countries at a clear competitive advantage than Kenya. Certain macroeconomic fundamentals that should be acting, as indicators of the levels of interest rates margins in Kenya seem not to be very indicative. A review of regional and global comparison of the trends in the determinants of interest rates is given below. 1.1.1. Profitability of banks Interest rates charged by commercial banks not only cover for the operation costs but also cover for the profits margin of the banks. Bank profitability is determined by the returns on assets and the returns on shareholders funds. According to the Banking Survey (2007), commercial banks’ returns on shareholders funds in Kenya are 37%. The returns on shareholders funds in Kenya are way above the returns earned in developed countries—for instance, Australia 15%, UK 20%, and Canada 17% and is even higher than the returns earned in some developing countries like South Africa at 22.5%, and Mauritius 14% (Competition commission for South Africa – CCSA, 2008). With these great disparities, bank profitability is probably one of the major drivers of the widening margins.
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
229
1.1.2. Corporate taxes Both explicit and implicit corporate taxes if high will significantly reduce the funds available for discretionary lending. A comparison of corporate taxes in Kenya and the other regional countries however shows no significant differences. South Africa levies a tax of 30% on the resident companies and 35% for the non-resident companies. In Singapore corporate taxes are 22% since 2003 for resident companies. This insignificance in differences may suggest that corporate taxes are not so important in explaining cross-country differences in the spreads. 1.1.3. Non-performing loans (NPLs) High levels of non-performing loans (NPLs) put a higher risk premium on subsequent loan advances. A specific provision for bad debts is then established to provide for an estimate of credit losses as soon as the recovery of a loan is identified as doubtful. These provisions deplete profits with the result that less money is left at the banks hands for discretionary lending leading to high lending rates and hence wider margins. The Kenyan banking sector has for a long time been burdened by the high levels of NPLs. The individual commercial bank’ share of the total NPL portfolio in Kenya is shown in Fig. 4 in Appendix B. 1.1.4. Cash and liquidity ratios (Monetary Policy operations) The CBK instruments – changes in cash and liquidity ratios, open market operation (OMO) and changes in the Central Bank rate (CBR) – are suppose to influence the direction of bank interest rates. The cash ratio in Kenya for instance has been reducing gradually from a high of 20% in 2004 to 5% in December 2008. This reduction was meant to improve bank liquidity and thereby reduce lending rates and increase deposit rates as more funds become available for lending and ultimately reduce interest spreads. Even with this significant reduction in cash ratio, interest margins have remained relatively unchanged and high. The CBR likewise had been reduced from a high of 9% in January 2009 to 7.5% by September 2009. This puts in doubt the significance of cash, liquidity ratios and the CBR (more generally monetary policy) in influencing the direction of interest rates in Kenya. Due to the decreased ratio, the banks instead of giving out more loans, invest more in treasury bills. 1.1.5. Banking sector market structure Even with the full liberalization of the financial sector in Kenya, competition in the banking sector has not become entrenched enough to bring about the desired reduction in the interest rates particularly the lending rates. In Kenya, the top five banks out of the 42 banking institutions control more than half (53%) of the total loan liabilities of the banking sector (Banking Survey, 2007, pp. 149). This leaves the other 37 banks to compete for the remaining 47% of the loan market. Further analyses of the total banking assets show bigger disparities. The largest five banks – Barclays Bank, Kenya Commercial Bank, Standard Chartered Bank (K) Ltd., the Co-operative Bank of Kenya and the CFC Bank – control more than half (51%) of the total banking assets (KBA, 2000). Together, the first 10 biggest banks in Kenya control almost three quarters of the banking industry assets at 72%. The individual share of each banks’ control of the total banking assets is given in Fig. 5 in Appendix B. The figure shows that out of the 42 banks, the bottom 32 banks control a meager 28% of the total banking assets. In light of this market structure, which is clearly concentrated at the top, competition in the banking sector in Kenya is not strong with the top banks not competing aggressively against each other. The structure of the banking industry in Kenya is however not any different from those of other developing countries. For instance according to the South African Competition Commission –
230
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
SACC (2007), most of the banking industry segments are dominated by what it calls the ‘Big four’—First National Bank (FNB), Standard Bank, ABSA and Nedbank. Their combined market share for a cluster of personal banking products (as at June 2007) accounted for approximately 95% of the total market share while the share of the remaining banks accounted for only 5% of the market share for the same period as shown in Fig. 6 in Appendix B. With these market structures, concentration levels and the formation of large financial conglomerates may develop to a point where an oligopolistic situation develops, characterized by informal cartel arrangements or informal market leadership. This may lead to inefficient allocation of loanable funds and is likely to drive interest rate spreads wider. 1.1.6. Bank operating costs and efficiency Another factor that is more institutional than fundamental is the commercial banks operating costs. The costs that commercial banks incur on salaries, rent, wages, water and rents are paid off from income earned from lending and other charges. These costs tend to increase and fall with the efficiency with which banks carry out their business and get translated into high or low lending rates. According to the Kenya Bankers Association (KBA, 2000), the highest component of the current cost structure are the operating costs at 8.79%. The other cost components, are the direct cost of funds at 7.44%, cost of liquidity at 1.25%, cost of holding cash at 0.25%, and the costs associated with the provisions for non-performing loans at 3.69% (KBA, 2000). An increase in these costs will translate into higher spreads as banks try to recover their costs. The next step in the analysis involves determining the relative importance of each of these determinants of interest rate spreads in each of the countries in the study and determining whether it can be concluded that some or all of the variables are important across the countries. 2. Literature review To develop the empirical model to be used in the analysis, we first review empirical literature on the determinants of financial sector efficiency. A number of studies have examined the relationship between interest rate margins, bank characteristics and other macroeconomic variables. Demirguc-Kunt, Laeven, and Levine (2003) in a study that examined the impacts of bank regulations, concentration, and national institutions on bank net interest margins in a panel of 72 countries, found out that bank concentration (market share) and regulatory restrictions are both positively related to the net interest margins. The authors noted that countries that restrict banks from engaging in non-traditional activities such as securities underwriting, real estates, owning non-financial and insurance firms, have margins that tend to be larger. Reserve requirements on the other hand are found to be positively related to interest rate margins. The study also finds that higher net interest rate margins tend to be associated with small banks, banks that hold a low fraction of liquid assets, banks that hold a relatively low amount of capital and banks without substantial income from fee based activities. Contrary to the findings of Demirguc-Kunt et al. (2003), Afanasieff, Lhacer, and Nakane (2001) finds that large banks in Brazil charge higher interest rate spreads. They interpreted this as exercise of market power by the large banks. Afanasieff et al. (2001) also found out that the ratio of non-interest bearing deposits to total operational assets affects positively the interest spread. Examining the factors behind the widening interest rate spread in Kenya following interest rate liberalization, Ndung’u and Ngugi (2000) found that, high implicit taxes, particularly the liquidity ratio and the cash ratio, increase the spread though the lending rates as the banks aim to maintain their profit margins. Liberalization is found to have widened interest margins, contrary
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
231
to the findings of Barajas, Steiner, and Salazar (1999) who finds that while overall spread did not reduce with financial liberalization in Columbia, the relevance of the different factors behind bank spreads were affected by liberalization. Ndung’u and Ngugi (2000), however, did not capture the influence of non-performing loans (credit risk) and market power which may be very important variables in explaining the financial market in/efficiency in Kenya. Demirguc-Kunt and Hiuzinga (1997) in a study that covered 80 countries found that there is a positive relationship between bank interest margins and the ratio of equity to lagged total assets. The study also found out that foreign banks realize relatively high net interest margins in relatively poor countries. Further, the study found that both interest margins and profitability increase with tax rates and that corporate taxes are passed on to bank customers to some degree. Angabazo (1997) in a study of the determinants of bank net interest margins for a sample of US banks found that default risk (ratio of net loan charge-offs to total loans), the opportunity cost of the non-interest bearing reserves (ratio of core capital to total assets), and management efficiency (ratio of earning assets to total assets) all increased bank interest margins. Just like Demirguc-Kunt et al. (2003), Angabazo (1997) finds that the ratio of liquid assets to total liabilities, a proxy for low liquidity risk, is inversely related to the bank interest margins. 3. Empirical framework and diagnostic tests This section gives the empirical framework followed in the study and the diagnostic test. 3.1. Empirical model This paper uses panel estimations for the cross-country analysis. In the model, a set of bankspecific characteristics and a set of macroeconomic variables as identified from the literature review are regressed against interest rate spread in the following panel regression yit = α + βXit + γZit + μit ,
μit ∼N(0, σit2 )
(1)
where μit = γi + λt + vit and γ i are unobservable individual effects, λt are unobservable time effects, vit is a stochastic disturbance term, yit is the interest spread for country i measured as the difference between the lending and the deposit rates in period t for (i = 1, . . ., N; t = 1, . . ., T) and N = 11 is the number of countries, T = 12 years is the number of years, Xit is a set of bank-specific variables in country i, in time t, Zit is a vector of macroeconomic variables in country i at time t, εit is the disturbance term, and α, β, γ are parameters to be estimated. The bank characteristic variables include: bank size, bank liquidity, default risk, bank liabilities, cost inefficiency, share turnover to capture alternative investments and bank concentration while the macroeconomic variables include treasury bill rates to capture fiscal policy pressures on interest rates. 3.2. Theoretical expectation on the variables Cost inefficiency is the ratio of total bank overheads to the gross income. From literature banks with higher operating costs to gross income ratio are expected to have higher interest rate spreads. The coefficient of the variable is therefore expected to be positive. Bank liquidity is defined as the ratio of total outstanding interbank debts to total bank deposits. The variable is expected to
232
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
be negatively related to the bank interest spreads. An increase in liquidity reduces the bank’s liquidity risks and liquidity premiums charged on loans. Default risk, which measures how risky a loan portfolio is, is expected to have a positive coefficient. Bank size, which are the total assets of bank i over the total assets of the banking industry is used in this study instead of concentration which was found to be stationary at levels. Increased bank size implies a more concentrated market. Increase in bank size will imply increased market power with a more concentrated banking sector. A concentrated banking industry from the structure-performance hypothesis will imply that the banks behave oligopolistically with the expectation that they have the leeway of fixing interest rates. Increased bank size will therefore widen the margins with an expected positive coefficient. A negative coefficient on the other hand would imply that the high concentration of banks leads to increased cost efficiency due to economies of scale. Increased bank liabilities will increase bank margins. The coefficient is expected to be positive. Share turnover at the stock markets indicate alternative investment in the market money other than putting money in the banks and treasury bills. It is expected that increased share turnover resulting from higher returns from the stock markets will pressurize banks to increase their deposit rates in order to attract money from the public in competition with the share market. This will have the effect of reducing interest rate margins. Fiscal policies just like the share markets act as competitors for funds with the banking sector. Increased returns from treasury bills will attract funds from the public starving banks of funds to give out in new loans. This would likely lead to shortage of loanable funds resulting in high lending rates and high interest margins. The coefficient on treasury bill rate is therefore expected to be positive. 3.3. Data and sample countries The countries included in the study are Kenya, Uganda, South Africa, Egypt, Japan, Germany, Malaysia, Poland, Korea, Botswana and USA. The choice of counties in the panel is based on regions and the level of financial sector development. It was projected that the study could cover more than the countries covered but due to data limitations, this was not possible. To achieve the objectives of this study, this paper uses panel cointegration technique with data from 1990 to 2007. This technique is chosen because it helps control for interest rate margin heterogeneity across countries. 3.4. Diagnostic tests Several diagnostic tests are conducted and the results reported in the following sub-sections. 3.4.1. Panel unit roots and panel cointegration According to Granger and Newbolt (1974), econometric estimation using non-stationary time series data often leads to spurious results (unless there is cointegration). We use the Im, Pesaran and Shin Test for panel unit roots (see Im, Pesaran, & Shin, 1995) to test for stationarity of the variables in the panel. The unit root test results reported in Table 1 in Appendix A fail to reject the null hypothesis that each of the series in all the cross-sections contains unit roots except for Bank concentration. This implies that all the series are non-stationary except concentration. We then use the residual based panel cointegration tests with both the Augmented Dickey Fuller (ADF) based test and the Phillips Perron (PP) based test both rejecting the null hypothesis of no cointegration. This means that the non-stationary series of the model are cointegrated. These findings allow us to proceed with the estimation of the model but we exclude bank
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
233
concentration from the estimations because it is integrated of a different order from the other variables. 3.4.2. Other panels tests We use the Chow Test to conduct poolability tests under the null hypothesis that all coefficients are equal against the alternative that not all are equal. The poolability test results indicate that we can decisively reject the null hypothesis of homogeneous coefficient. This means that the slopes of the different cross-sections vary. We also conduct exogeneity tests using the Hausman Test. This test is equivalent to testing for random effects against a fixed effects model. The test results reject the null hypothesis of random effects implying that we use the fixed effects model. 3.5. Estimation technique Test on the cross-section disturbances revealed that the disturbances are correlated due to different similarities in the countries used in the study. To solve this problem, this paper uses Zellner (1962)’s seemingly unrelated regression (SUR). Contemporaneous cross-country correlation may be caused by being members of the same regional trading bodies, the time of liberalization, same bank characteristics across the countries by the fact that the banks are subsidiaries of the same parent like Barclays bank, general state of the economy in the different countries, the political environment, etc. The estimation results of the SUR model are reported in Table 2 in Appendix A 4. Estimation results The results given in Table 2 in Appendix A shows that the impact of default risk is positive and statistically significant for Kenya. Default risk as used in this study, is an index that measures how risky the loan portfolios are. This result implies that as default risk increases, the banks increase their default premium on new loans and this is reflected on higher lending rates which increases interest margins. The high default risk and low quality of the loans could be attributed to the high non-performing loans portfolio in Kenya. The only countries where default risk is not a problem are South Africa and Poland. For all the other countries in the study, default risk is a major determinant of the interest spreads in the respective counties. The magnitude of impact of default risk on interest spreads for Kenya is higher than those of all the other countries except Uganda and Botswana. Uganda and Botswana’s spreads are however wider than Kenya’s. It can be inferred therefore that countries with higher default risk experience wider interest margins and therefore high default risk is one of the major reasons why interest margins are wider in Kenya compared to the other countries. Bank liquidity variable, which is the ratio of outstanding interbank debt to total deposits, was expected to have a negative coefficient. The variable is instead found to be positive and statistically significant (at the 10% level) for Kenya. The positive coefficient implies that an increase in bank liquidity widens the margins in Kenya. This could imply that, as banks’ liquidity positions improve, they invest the excess liquidity in other securities like treasury bill rates instead of using it to increase lending. If the extra liquidity were used in lending activities, the lending rates would go down reducing the margins, but this is not the case. This could imply that treasury bill rates offered by the government are higher and more attractive than the lending rates. Therefore, the banks obtain higher returns from investing in Treasury bills than giving out loans.
234
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
The bank size which in this study indicates market power is found to be positive for Kenya, Japan, Egypt and Botswana with the coefficients all statistically different from zero. This implies that bank concentration have significantly widened the interest margins in these countries. The other countries including USA, Korea, S. Africa, Uganda, Poland, Germany and Malaysia all have negative coefficients. This fact may be indicative of lowering of margins as bank concentration increases probably due to increased efficiency and economies of scale. However, the coefficients are not significant for Malaysia and Poland implying that bank size is not important in explaining the margins in those countries. The coefficients for Uganda, South Africa and USA are all significant at 10% significant level. Among the countries with positive coefficients, the magnitude of the coefficients is highest in Kenya; higher than both Egypt and Japan. This implies that countries with a more concentrated market structure also experience higher interest margins. Market concentration is therefore one of the factors that explain why Kenya’s margins are higher (financial sector is more inefficient) than some of the other countries in the study. Botswana on the other hand has a bank size coefficient that is higher than Kenya’s implying that Botswana’s financial market is more concentrated than Kenya’s. This result is not surprising since Botswana’s financial sector is more inefficient than Kenya’s (wider spreads than Kenya). The cost inefficiency in the banking industry reflects the degree of inefficiency in delivering banking services. The coefficients for all the countries are positive indicating that increased cost inefficiency widens the margins in the respective countries. Cost inefficiency is important in explaining interest rate margins in Kenya, USA, Korea, Germany and Japan with the coefficient for Poland significant at 10% level. Compared to these other countries Kenya has the highest cost inefficiency impact on the margins at 69%. It is not surprising that these countries also have lower spreads than Kenya. This implies that if a country is less cost-inefficient, it will experience wider margins. Inefficiency in delivering financial services therefore is another of the major reasons why Kenya’s financial sector is more inefficient compared to the other countries. The results further show that increased activity and returns from the stock markets improves the efficiency (reduces inefficiency) of the financial markets in all the countries under study as it increases competition for deposits with the banks. However, the coefficient is not significant for Egypt, Malaysia and Botswana. But since the impact of the stock market activity in reducing financial market inefficiency is lower in the USA for instance than in Kenya, while the spreads for USA are at the same time lower than Kenya’s, it cannot be concluded that a more active stock market is the reason why USA’s margins are narrower. If that was the case, the reduction in inefficiency due to the improved activity in the stock market would be higher in the USA than Kenya. Therefore, stock market activity is an important factor in reducing financial sector inefficiency, but is not one of the reasons why the financial markets in Kenya are more inefficient than the other countries. Except for Kenya and Egypt, which have statistically insignificant coefficients for treasury bill rates, all the other countries used in the study indicate that treasury bill rates, which are an alternative investment window for public funds are important in explaining the bank interest rate margins in the respective countries. The coefficients of Japan, Germany, Malaysia and Poland are significant at the 10% level. In general, the results show that in terms of magnitude, cost inefficiency is the major contributor to financial market inefficiency in Kenya followed by financial market structure, and default risk in that order. Stock market activity is an important factor in reducing financial sector inefficiency in Kenya, but is not one of the reasons why the financial markets in Kenya are more inefficient
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
235
than the other countries. Fiscal pressures on the other hand are found to be insignificant in driving financial market in/efficiency. The beauty of assuming different coefficients in the panel as can be seen from the above results is that the factors that contribute to market inefficiency in one country are not necessarily the same as the factors that lead to market inefficiency in the other countries and the magnitude of the impact of a variable differs from one country to the other. 5. Conclusions Widening interest rate margins, which is the indicator of financial sector inefficiency in this study is hypothesized to be detrimental for savings mobilization and stifles investment growth. Wide interest margins as witnessed in Kenya are therefore a sign of a repressed and inefficient financial sector. This paper uses panel cointegration and data from 1990 to 2007 to analyze the determinants of the efficiency of the financial sector intermediation process in 11 countries with a view to recommending policy options for reducing the spreads and improving the financial market efficiency in Kenya. The countries are selected from the African region and other developed countries and includes Kenya, South Africa, Uganda, Egypt, Japan, Germany, Malaysia, Poland, Korea, Botswana and USA. The results show that cost inefficiency is important in explaining interest rate margins in Kenya, USA, Korea, Germany and Japan and Poland with the highest impact of cost inefficiency recorded in Kenya at 69%. It is important to note that these other countries have lower spreads than Kenya. It can therefore be inferred from this result that the other countries have lower margins than Kenya because the impact of cost inefficiency on their margins is not as huge as Kenya’s. In terms of the magnitude, cost inefficiency is found to be the major determinant of wide margins (market inefficiency) in Kenya. The results also show that the impact of default risk is positive and statistically significant for Kenya. The only countries where default risk is not a problem are South Africa and Poland. Important is the finding that the magnitude of impact of default risk on interest spreads for Kenya is higher than those of all the other countries except Uganda and Botswana. All the other countries except Uganda and Botswana have lower spreads than Kenya. It can be inferred again from this result therefore that default risk is one of the major contributors to the financial market inefficiency in Kenya compared to the other countries. This is because the magnitude of impact is higher in Kenya and therefore margins are higher in Kenya compared to the other countries. Default risk is found to be the second most major contributor to inefficient financial markets in Kenya. The third major factor that is important in explaining the financial market inefficiency in Kenya is bank size (financial market structure). Among the countries with positive coefficients, Kenya reports the second highest magnitude of the impact of financial market structure on market inefficiency. Botswana has the highest coefficient on bank size among the countries with positive coefficients but Botswana also has higher inefficiency (wider spreads) than Kenya. This implies that countries with higher bank size (a less competitive market) would experience wider interest margins than countries with more competitive structures. The result here implies that the nature of the financial market structure in Kenya is one of the major factors that explains the very high interest margins (inefficient financial sector) in Kenya compared to the other countries in the study except Botswana. Stock market activity is found to be an important factor in reducing financial sector inefficiency in Kenya but does not account for cross-country differences in market inefficiency between Kenya
236
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
and the other countries. Fiscal pressures on the other hand are found to be insignificant in the Kenyan panel and therefore do not explain the differences in market efficiency between Kenya and the other countries. The results show that the factors that lead to market inefficiency in one country are not necessarily the same as the factors that lead to market inefficiency in the other countries. 5.1. Policy recommendations The findings show that one the main reason why interest rate spreads are high in Kenya compared to the other countries are high cost inefficiencies. To reduce bank operating costs, the banks need to put in place technologically innovative ways of doing business. These include the use of phone banking and e-banking. The recent initiatives towards automation of bank services are a step towards in the right direction, but more still needs to be done. For instance, most teller machines still cannot be used to make deposits but only withdrawals. The second major reason why financial markets in Kenya are more inefficient compared to other countries is high default risk. To make the default risk in the credit market low, banks and the government through the central bank must establish a credit reference system to identify customers who are not credit worthy. There is a credit reference bureau in Kenya, but it lacks the regulatory environment and the support of both the government and the banking sector. The results further show that financial market structure is one of the major contributors to inefficiency in the financial sector in Kenya. The CBK has closed doors to licensing of new banks over the last few years. In this period only K-Rep Bank and Equity bank has been licensed to operate as mainstream banks. Closing the door to new entrants is a sure way of blocking competition from new entrants. The government through the CBK must therefore come in to license more new banks for increased competition. Appendix A. See Tables A.1 and A.2.
Table A.1 Summary of panel unit root tests. Variable
Margins Concentration Cost inefficiency Bank size Bank liabilities Bank liquidity Default risk Share turnover Tbill rates a
Null: unit root
Conclusion
ADF
PP
−0.37 −4.00a 0.26 0.21 1.31 1.03 2.72 4.64 −0.08
1.17 −2.21a −4.65 0.61 0.79 2.50 2.99 10.13 −0.92
Rejection of null hypothesis (of non-stationarity) at 5% with a test statistic of −1.645.
Non-stationary Stationary Non-stationary Non-stationary Non-stationary Non-stationary Non-stationary Non-stationary Non-stationary
Table A.2 SUR results.
Japan Germany Egypt Malaysia Poland Uganda South Africa Korea Botswana USA
Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic
Bank liquidity
Bank liability
Default risk
Cost inefficiency
Share turmover
Treasury bills
0.61 7.25 0.09 16.55 −0.01 −2.55 0.03 1.92 −0.01 −0.80 −0.01 −0.38 −1.51 −1.42 −0.01 −1.34 −0.12 −3.99 2.78 6.63 −0.02 −1.20
0.09 1.75 −0.03 −4.54 −0.004 −1.34 −0.19 −6.36 −0.07 −1.91 −0.65 −2.79 −1.74 −1.30 0.02 0.64 0.05 6.81 0.30 2.47 0.43 2.78
0.31 0.08 0.001 0.25 0.07 3.13 0.13 5.71 0.02 1.85 0.52 1.71 2.85 2.15 0.68 2.67 0.02 5.39 0.23 3.37 0.45 2.82
0.64 10.36 0.11 8.02 0.04 2.47 0.04 4.19 0.05 3.69 0.08 0.06 4.27 1.69 0.01 0.07 0.07 2.18 3.31 7.05 0.03 3.17
0.69 2.10 0.18 2.98 0.35 3.79 0.38 0.98 0.03 0.54 0.24 1.36 0.03 0.16 0.12 0.39 0.12 5.30 0.19 0.60 0.28 2.08
−0.19 −2.22 −0.003 −2.14 −0.001 −1.26 −0.0003 −0.10 −0.001 −0.32 −0.01 −1.48 −4825.51 −1.15 −0.17 −3.14 −0.002 −7.13 −0.01 −0.48 −0.002 −1.92
−0.0004 −0.06 −0.02 −1.15 −0.03 −1.38 −0.01 −0.68 −0.11 −1.44 −0.08 −1.24 −0.11 −2.35 −0.68 −2.99 −0.06 −7.94 −0.42 −5.25 −0.12 −3.87
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
Kenya
Banking size
237
238
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
Appendix B. See Figs. B.1–B.6. 40 35
percent
30 25 20 15 10
2007
2008
2006
2004
2005
2002
2003
2001
2000
1999
1997
1998
1996
1995
1993
1994
1992
1991
0
1990
5
Year Deposit rate
Lending rates
Source: International Financial statistics (IFS)-February 2008
Fig. B.1. Interest rate Margins in Kenya (1990–2007). International Financial statistics (IFS) – February 2008. 25
percent
20 15 10
2006
2004
2005
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
0
1990
5
Year Kenya
Korea
Singapore
South Africa
United States
Source: IFS-February 2008
Fig. B.2. Interest rate margins in Kenya and some selected countries of the world. IFS – February 2008. 30
percent
25 20 15 10 5 2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
0
Year Kenya
Egypt
South Africa
Tanzania
Uganda
Source IFS- February 2008
Fig. B.3. Interest margins between Kenya and other regional countries. IFS – February 2008.
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240 Guardian Bank 1% NIC bank 1%
AEBS Bank 3%
Standard Chatered Bank 3%
HFCK 4%
Others 13%
Barclays Bank of Kenya 9% Kenya Commercial Bank 12%
National Bank of Kenya 36%
Co-operative Bank of Kenya 18%
Source: Banking Survey (2007)
Fig. B.4. Non-performing loans in Kenya. Banking Survey (2007).
Fig. B.5. Market share of Kenyan Banks. Banking Survey (2007, pp. 146). Nedbank 9%
Other 5%
Standard Nabk 30%
FNB 20%
ABSA 36%
Source: SACC (2007) Fig. B.6. South African Banks Market Shares (June 2007). SACC (2007).
239
240
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
References Afanasieff, T. S., Lhacer, P. M., & Nakane, M. I. (2001). The determinants of bank interest spread in Brazil. In Proceedings of the XXIX Encontro Nacional da Economia, Salvador ANPEC. Angabazo, L. (1997). Commercial bank net interest margins, default risk, interest-rate risk, and off-balance sheet banking. Journal of Banking and Finance, 21, 55–87. Barajas, A, Steiner, R., & Salazar, N. (1999). Interest spreads in banking in Colombia. IMF Staff Papers, 46(2), 196–225. CCSA. (2008). Banking Industry Survey. Pretoria, South Africa: CCSA. Demirguc-Kunt, A., Laeven, L., & Levine, R. (2003) The impact of bank regulations, concentration, and institutions on bank margins. Policy Research Working Paper Series 3030, The World Bank. Demirguc-Kunt, & Hiuzinga, H. (1997). Determinants of commercial bank interest margins and profitability: Some international evidence. World Bank working paper No. 1900, Washington DC. Granger, C. W. J., & Newbolt, P. (1974). Spurious regressions in econometrics. Journal of Econometrics, 2, 111–120. Im, K. S., Pesaran, M. H., & Shin, Y. (1995). Testing for unit roots in panel data. Department of Applied Economics, University of Cambridge. Manuscript. Kenya Bankers Association (KBA) (2000). A statement by KBA to the Parliamentary Finance Committee, on the Central Bank (Amendment) Bill, 2000. Ndung’u, S. N., & Ngugi, R. W. (2000). Banking sector interest rate spread in Kenya. KIPPRA paper DP/05/2000. Zellner, A. (1962). An efficient method of estimating seemingly unrelated regression equations and tests for aggregation bias. Journal of the American Statistical Association, 57, 348–368.
Journal of Policy Modeling 33 (2011) 226–240
Efficiency of financial market intermediation in Kenya: A comparative analysis Jacob Oduor a,∗ , Stephen Karingi b , Stephen Mwaura c a
Kenyatta University and the Kenya Institute for Public Policy Research and Analysis, Kenya, P.O. Box 56445-00200 Nairobi, Kenya b United Nations Economic Commission for Africa, Addis-Ababa, Ethiopia c Central Bank of Kenya, Kenya Received 1 March 2010; received in revised form 1 July 2010; accepted 1 August 2010 Available online 21 September 2010
Abstract Wide interest margins as witnessed in Kenya are a sign of a repressed and inefficient financial sector. This paper carries out a cross-country analysis of the determinants of financial market efficiency using panel cointegration with a view to recommending policy options for improving the efficiency of the financial sector intermediation process in Kenya. The study finds that the major contributors to the differences in financial sector inefficiency in Kenya compared to the other countries in the study are high bank operating costs, default risk and financial market structure. The study recommends, among other measures, that the government through the Central Bank need to collaborate with the commercial banks and establish a working credit reference bureau to enable easy identification of credit worthy customers in order to reduce default risk; there is also need by the central bank to license more new banks to increase competition and reduce bank concentration. The study also recommends increased use of technology including phone-banking and e-banking to reduce operation costs of the banks. The paper concludes that contrary to the findings from other cross-country analysis, the factors that lead to financial market in/efficiency varies from one country to the other. © 2010 Society for Policy Modeling. Published by Elsevier Inc. All rights reserved. JEL classification: G14; G21; C33 Keywords: Financial market efficiency; Panel cointegration
∗
Corresponding author. Tel.: +254 722978186. E-mail addresses: [email protected], [email protected] (J. Oduor).
0161-8938/$ – see front matter © 2010 Society for Policy Modeling. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.jpolmod.2010.09.003
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
227
1. Background The ability of the financial market to efficiently allocate and reallocate loanable funds is critical for economic growth. When financial markets are efficient, the resultant competition among the financial institutions and reduced transaction and operation costs will imply higher deposit rates and lower lending rates as the financial institutions compete to attract borrowers and savers. If competition for instance is lacking in the market, there is a possibility that lending rates may be higher than the free market average and deposit rates will be lower than the free market rates. Would-be depositors will therefore not save as much. They may instead opt to consume and not save. This stifles savings mobilization and reduces funds available for investments and leads to misallocation of resources in the economy. On the other hand, investors will shy away from borrowing as the lending rates rise. Since the main role of financial institutions is to intermediate between the borrowers and lenders in the market, high lending rates and low deposit rates, and by implication, wide interest rate spreads, are indications of inefficient financial sector intermediation process. Interest rate spread is the difference between lending rates and deposit rates and has been used severally in literature to indicate the efficiency of the financial sector intermediation process (see for instance Demirguc-Kunt & Hiuzinga, 1997). Widening spreads therefore indicate increasingly inefficient financial market. Discussions on interest rate spreads in Kenya have dominated many forums on economics and business in the recent past. The point is clear—lending interest rates in Kenya though have been falling from 2003 still remains very high while deposit rates are very low. This, as can be seen in Fig. 1 in Appendix B, has left interest rate margins relatively high and unchanged. At the gist of this discussion is the concern about the consequences that such high interest rates spreads have on economic performance of the country and on the welfare of the households and firms who must borrow to invest or save to consume in future. This concern puts to question the ability of the financial sector in Kenya to mediate efficiently between borrowers and lenders in the economy. The wide margins as witnessed in Kenya, is a sign of financial sector repression, which compromises the financial sector’s role in the intermediation process of deposit mobilization and credit creation. The other concern is the implication of such wide margins on Kenya’s global competitiveness. The commercial banks in Kenya levy over 12 percentage points spread between the deposit and lending rates as opposed for instance to 3.7% in South Africa, 3% in Singapore and 1% in Korea (Banking Survey, 2007). These differences jeopardize Kenya’s international competitiveness as international competitors are at an advantage of at least 4% points in their cost of production against their Kenyan counterparts. A comparison of some of the fundamental determinants of interest rates with those of other countries shows no significant differences. For instance, banking sector concentrations in South Africa and Egypt are almost similar to Kenya’s yet interest margins in the two countries are much lower than in Kenya. The questions that come to mind therefore are ‘why are interest margins so wide in Kenya compared to the other countries and what are the driving factors?’ In light of these concerns, it is important therefore to examine the relative importance of the determinants of interest rate margins in Kenya vis a vis those of other countries to determine the factors that explain the high margins and are specific to Kenya and not the other countries. Studies on interest rate margins in Kenya like Ndung’u and Ngugi (2000) have not analyzed cross-country differences in the financial sector determinants of interest rates spreads to establish whether the determinants significantly differ across countries. In addition, Ndung’u and Ngugi (2000) did not capture the influence of non-performing loans (credit risk), market power or structure, and the transaction costs which are very important variables in explaining the financial market efficiency. This study, in addition to filling this knowledge gap, attempts to inform policy
228
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
on the major focus areas in line with the significance of the fundamentals in the other countries as compared to their significance in Kenya. To analyze the main determinants of interest rate spreads as an indicator of financial sector efficiency, we use panel cointegration method with data from 1990 to 2007 for 11 countries. The results show that the major contributors to the financial sector inefficiency in Kenya compared to the other countries in the study are high operating costs (cost inefficiency), default risk or level of non-performing loans and bank size or bank concentration. In terms of magnitude, cost inefficiency is the major contributor to financial market inefficiency in Kenya followed by financial market structure, and default risk in that order. Further the findings show that the factors that lead to market inefficiency in one country are not necessarily the same as the factors that lead to market inefficiency in the other countries. The study recommends, among other measures, that the government through the Central Bank and the commercial banks need to collaborate and establish a working credit reference bureau to identify easily credit worthy customers in order to reduce default risk and also stresses on the need by the central bank to license more new banks to increase competition and reduce bank concentration. There is also need to stress on increased use of technology including phone-banking and e-banking to reduce operation costs of the banks. The rest of this paper is organized as follows; the remainder of Section 1 gives the comparisons of interest rate spreads and its determinants in Kenya with those of other countries. Section 2 gives the literature review, Section 3 gives the empirical framework adopted and the diagnostic tests while Section 4 discusses the empirical results. Section 5 concludes and gives policy recommendations suggested. 1.1. Comparisons of some of the fundamental determinants of interest margins Comparisons of Kenyan margins with those of some of Kenya’s strongest competitors in the export markets reveals great disparities. The comparisons show quite lower margins for the other countries including among others; Korea, and Singapore (see Fig. 2 in Appendix B) and even regionally including South Africa and Egypt (see Fig. 3 in Appendix B). Whereas Kenya had average interest rate margins of 11.5% in 2007, Kenya’s main competitors South Africa have 3.7% and Egypt on the other hand has interest margins of 3.2%. This puts these countries at a clear competitive advantage than Kenya. Certain macroeconomic fundamentals that should be acting, as indicators of the levels of interest rates margins in Kenya seem not to be very indicative. A review of regional and global comparison of the trends in the determinants of interest rates is given below. 1.1.1. Profitability of banks Interest rates charged by commercial banks not only cover for the operation costs but also cover for the profits margin of the banks. Bank profitability is determined by the returns on assets and the returns on shareholders funds. According to the Banking Survey (2007), commercial banks’ returns on shareholders funds in Kenya are 37%. The returns on shareholders funds in Kenya are way above the returns earned in developed countries—for instance, Australia 15%, UK 20%, and Canada 17% and is even higher than the returns earned in some developing countries like South Africa at 22.5%, and Mauritius 14% (Competition commission for South Africa – CCSA, 2008). With these great disparities, bank profitability is probably one of the major drivers of the widening margins.
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
229
1.1.2. Corporate taxes Both explicit and implicit corporate taxes if high will significantly reduce the funds available for discretionary lending. A comparison of corporate taxes in Kenya and the other regional countries however shows no significant differences. South Africa levies a tax of 30% on the resident companies and 35% for the non-resident companies. In Singapore corporate taxes are 22% since 2003 for resident companies. This insignificance in differences may suggest that corporate taxes are not so important in explaining cross-country differences in the spreads. 1.1.3. Non-performing loans (NPLs) High levels of non-performing loans (NPLs) put a higher risk premium on subsequent loan advances. A specific provision for bad debts is then established to provide for an estimate of credit losses as soon as the recovery of a loan is identified as doubtful. These provisions deplete profits with the result that less money is left at the banks hands for discretionary lending leading to high lending rates and hence wider margins. The Kenyan banking sector has for a long time been burdened by the high levels of NPLs. The individual commercial bank’ share of the total NPL portfolio in Kenya is shown in Fig. 4 in Appendix B. 1.1.4. Cash and liquidity ratios (Monetary Policy operations) The CBK instruments – changes in cash and liquidity ratios, open market operation (OMO) and changes in the Central Bank rate (CBR) – are suppose to influence the direction of bank interest rates. The cash ratio in Kenya for instance has been reducing gradually from a high of 20% in 2004 to 5% in December 2008. This reduction was meant to improve bank liquidity and thereby reduce lending rates and increase deposit rates as more funds become available for lending and ultimately reduce interest spreads. Even with this significant reduction in cash ratio, interest margins have remained relatively unchanged and high. The CBR likewise had been reduced from a high of 9% in January 2009 to 7.5% by September 2009. This puts in doubt the significance of cash, liquidity ratios and the CBR (more generally monetary policy) in influencing the direction of interest rates in Kenya. Due to the decreased ratio, the banks instead of giving out more loans, invest more in treasury bills. 1.1.5. Banking sector market structure Even with the full liberalization of the financial sector in Kenya, competition in the banking sector has not become entrenched enough to bring about the desired reduction in the interest rates particularly the lending rates. In Kenya, the top five banks out of the 42 banking institutions control more than half (53%) of the total loan liabilities of the banking sector (Banking Survey, 2007, pp. 149). This leaves the other 37 banks to compete for the remaining 47% of the loan market. Further analyses of the total banking assets show bigger disparities. The largest five banks – Barclays Bank, Kenya Commercial Bank, Standard Chartered Bank (K) Ltd., the Co-operative Bank of Kenya and the CFC Bank – control more than half (51%) of the total banking assets (KBA, 2000). Together, the first 10 biggest banks in Kenya control almost three quarters of the banking industry assets at 72%. The individual share of each banks’ control of the total banking assets is given in Fig. 5 in Appendix B. The figure shows that out of the 42 banks, the bottom 32 banks control a meager 28% of the total banking assets. In light of this market structure, which is clearly concentrated at the top, competition in the banking sector in Kenya is not strong with the top banks not competing aggressively against each other. The structure of the banking industry in Kenya is however not any different from those of other developing countries. For instance according to the South African Competition Commission –
230
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
SACC (2007), most of the banking industry segments are dominated by what it calls the ‘Big four’—First National Bank (FNB), Standard Bank, ABSA and Nedbank. Their combined market share for a cluster of personal banking products (as at June 2007) accounted for approximately 95% of the total market share while the share of the remaining banks accounted for only 5% of the market share for the same period as shown in Fig. 6 in Appendix B. With these market structures, concentration levels and the formation of large financial conglomerates may develop to a point where an oligopolistic situation develops, characterized by informal cartel arrangements or informal market leadership. This may lead to inefficient allocation of loanable funds and is likely to drive interest rate spreads wider. 1.1.6. Bank operating costs and efficiency Another factor that is more institutional than fundamental is the commercial banks operating costs. The costs that commercial banks incur on salaries, rent, wages, water and rents are paid off from income earned from lending and other charges. These costs tend to increase and fall with the efficiency with which banks carry out their business and get translated into high or low lending rates. According to the Kenya Bankers Association (KBA, 2000), the highest component of the current cost structure are the operating costs at 8.79%. The other cost components, are the direct cost of funds at 7.44%, cost of liquidity at 1.25%, cost of holding cash at 0.25%, and the costs associated with the provisions for non-performing loans at 3.69% (KBA, 2000). An increase in these costs will translate into higher spreads as banks try to recover their costs. The next step in the analysis involves determining the relative importance of each of these determinants of interest rate spreads in each of the countries in the study and determining whether it can be concluded that some or all of the variables are important across the countries. 2. Literature review To develop the empirical model to be used in the analysis, we first review empirical literature on the determinants of financial sector efficiency. A number of studies have examined the relationship between interest rate margins, bank characteristics and other macroeconomic variables. Demirguc-Kunt, Laeven, and Levine (2003) in a study that examined the impacts of bank regulations, concentration, and national institutions on bank net interest margins in a panel of 72 countries, found out that bank concentration (market share) and regulatory restrictions are both positively related to the net interest margins. The authors noted that countries that restrict banks from engaging in non-traditional activities such as securities underwriting, real estates, owning non-financial and insurance firms, have margins that tend to be larger. Reserve requirements on the other hand are found to be positively related to interest rate margins. The study also finds that higher net interest rate margins tend to be associated with small banks, banks that hold a low fraction of liquid assets, banks that hold a relatively low amount of capital and banks without substantial income from fee based activities. Contrary to the findings of Demirguc-Kunt et al. (2003), Afanasieff, Lhacer, and Nakane (2001) finds that large banks in Brazil charge higher interest rate spreads. They interpreted this as exercise of market power by the large banks. Afanasieff et al. (2001) also found out that the ratio of non-interest bearing deposits to total operational assets affects positively the interest spread. Examining the factors behind the widening interest rate spread in Kenya following interest rate liberalization, Ndung’u and Ngugi (2000) found that, high implicit taxes, particularly the liquidity ratio and the cash ratio, increase the spread though the lending rates as the banks aim to maintain their profit margins. Liberalization is found to have widened interest margins, contrary
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
231
to the findings of Barajas, Steiner, and Salazar (1999) who finds that while overall spread did not reduce with financial liberalization in Columbia, the relevance of the different factors behind bank spreads were affected by liberalization. Ndung’u and Ngugi (2000), however, did not capture the influence of non-performing loans (credit risk) and market power which may be very important variables in explaining the financial market in/efficiency in Kenya. Demirguc-Kunt and Hiuzinga (1997) in a study that covered 80 countries found that there is a positive relationship between bank interest margins and the ratio of equity to lagged total assets. The study also found out that foreign banks realize relatively high net interest margins in relatively poor countries. Further, the study found that both interest margins and profitability increase with tax rates and that corporate taxes are passed on to bank customers to some degree. Angabazo (1997) in a study of the determinants of bank net interest margins for a sample of US banks found that default risk (ratio of net loan charge-offs to total loans), the opportunity cost of the non-interest bearing reserves (ratio of core capital to total assets), and management efficiency (ratio of earning assets to total assets) all increased bank interest margins. Just like Demirguc-Kunt et al. (2003), Angabazo (1997) finds that the ratio of liquid assets to total liabilities, a proxy for low liquidity risk, is inversely related to the bank interest margins. 3. Empirical framework and diagnostic tests This section gives the empirical framework followed in the study and the diagnostic test. 3.1. Empirical model This paper uses panel estimations for the cross-country analysis. In the model, a set of bankspecific characteristics and a set of macroeconomic variables as identified from the literature review are regressed against interest rate spread in the following panel regression yit = α + βXit + γZit + μit ,
μit ∼N(0, σit2 )
(1)
where μit = γi + λt + vit and γ i are unobservable individual effects, λt are unobservable time effects, vit is a stochastic disturbance term, yit is the interest spread for country i measured as the difference between the lending and the deposit rates in period t for (i = 1, . . ., N; t = 1, . . ., T) and N = 11 is the number of countries, T = 12 years is the number of years, Xit is a set of bank-specific variables in country i, in time t, Zit is a vector of macroeconomic variables in country i at time t, εit is the disturbance term, and α, β, γ are parameters to be estimated. The bank characteristic variables include: bank size, bank liquidity, default risk, bank liabilities, cost inefficiency, share turnover to capture alternative investments and bank concentration while the macroeconomic variables include treasury bill rates to capture fiscal policy pressures on interest rates. 3.2. Theoretical expectation on the variables Cost inefficiency is the ratio of total bank overheads to the gross income. From literature banks with higher operating costs to gross income ratio are expected to have higher interest rate spreads. The coefficient of the variable is therefore expected to be positive. Bank liquidity is defined as the ratio of total outstanding interbank debts to total bank deposits. The variable is expected to
232
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
be negatively related to the bank interest spreads. An increase in liquidity reduces the bank’s liquidity risks and liquidity premiums charged on loans. Default risk, which measures how risky a loan portfolio is, is expected to have a positive coefficient. Bank size, which are the total assets of bank i over the total assets of the banking industry is used in this study instead of concentration which was found to be stationary at levels. Increased bank size implies a more concentrated market. Increase in bank size will imply increased market power with a more concentrated banking sector. A concentrated banking industry from the structure-performance hypothesis will imply that the banks behave oligopolistically with the expectation that they have the leeway of fixing interest rates. Increased bank size will therefore widen the margins with an expected positive coefficient. A negative coefficient on the other hand would imply that the high concentration of banks leads to increased cost efficiency due to economies of scale. Increased bank liabilities will increase bank margins. The coefficient is expected to be positive. Share turnover at the stock markets indicate alternative investment in the market money other than putting money in the banks and treasury bills. It is expected that increased share turnover resulting from higher returns from the stock markets will pressurize banks to increase their deposit rates in order to attract money from the public in competition with the share market. This will have the effect of reducing interest rate margins. Fiscal policies just like the share markets act as competitors for funds with the banking sector. Increased returns from treasury bills will attract funds from the public starving banks of funds to give out in new loans. This would likely lead to shortage of loanable funds resulting in high lending rates and high interest margins. The coefficient on treasury bill rate is therefore expected to be positive. 3.3. Data and sample countries The countries included in the study are Kenya, Uganda, South Africa, Egypt, Japan, Germany, Malaysia, Poland, Korea, Botswana and USA. The choice of counties in the panel is based on regions and the level of financial sector development. It was projected that the study could cover more than the countries covered but due to data limitations, this was not possible. To achieve the objectives of this study, this paper uses panel cointegration technique with data from 1990 to 2007. This technique is chosen because it helps control for interest rate margin heterogeneity across countries. 3.4. Diagnostic tests Several diagnostic tests are conducted and the results reported in the following sub-sections. 3.4.1. Panel unit roots and panel cointegration According to Granger and Newbolt (1974), econometric estimation using non-stationary time series data often leads to spurious results (unless there is cointegration). We use the Im, Pesaran and Shin Test for panel unit roots (see Im, Pesaran, & Shin, 1995) to test for stationarity of the variables in the panel. The unit root test results reported in Table 1 in Appendix A fail to reject the null hypothesis that each of the series in all the cross-sections contains unit roots except for Bank concentration. This implies that all the series are non-stationary except concentration. We then use the residual based panel cointegration tests with both the Augmented Dickey Fuller (ADF) based test and the Phillips Perron (PP) based test both rejecting the null hypothesis of no cointegration. This means that the non-stationary series of the model are cointegrated. These findings allow us to proceed with the estimation of the model but we exclude bank
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
233
concentration from the estimations because it is integrated of a different order from the other variables. 3.4.2. Other panels tests We use the Chow Test to conduct poolability tests under the null hypothesis that all coefficients are equal against the alternative that not all are equal. The poolability test results indicate that we can decisively reject the null hypothesis of homogeneous coefficient. This means that the slopes of the different cross-sections vary. We also conduct exogeneity tests using the Hausman Test. This test is equivalent to testing for random effects against a fixed effects model. The test results reject the null hypothesis of random effects implying that we use the fixed effects model. 3.5. Estimation technique Test on the cross-section disturbances revealed that the disturbances are correlated due to different similarities in the countries used in the study. To solve this problem, this paper uses Zellner (1962)’s seemingly unrelated regression (SUR). Contemporaneous cross-country correlation may be caused by being members of the same regional trading bodies, the time of liberalization, same bank characteristics across the countries by the fact that the banks are subsidiaries of the same parent like Barclays bank, general state of the economy in the different countries, the political environment, etc. The estimation results of the SUR model are reported in Table 2 in Appendix A 4. Estimation results The results given in Table 2 in Appendix A shows that the impact of default risk is positive and statistically significant for Kenya. Default risk as used in this study, is an index that measures how risky the loan portfolios are. This result implies that as default risk increases, the banks increase their default premium on new loans and this is reflected on higher lending rates which increases interest margins. The high default risk and low quality of the loans could be attributed to the high non-performing loans portfolio in Kenya. The only countries where default risk is not a problem are South Africa and Poland. For all the other countries in the study, default risk is a major determinant of the interest spreads in the respective counties. The magnitude of impact of default risk on interest spreads for Kenya is higher than those of all the other countries except Uganda and Botswana. Uganda and Botswana’s spreads are however wider than Kenya’s. It can be inferred therefore that countries with higher default risk experience wider interest margins and therefore high default risk is one of the major reasons why interest margins are wider in Kenya compared to the other countries. Bank liquidity variable, which is the ratio of outstanding interbank debt to total deposits, was expected to have a negative coefficient. The variable is instead found to be positive and statistically significant (at the 10% level) for Kenya. The positive coefficient implies that an increase in bank liquidity widens the margins in Kenya. This could imply that, as banks’ liquidity positions improve, they invest the excess liquidity in other securities like treasury bill rates instead of using it to increase lending. If the extra liquidity were used in lending activities, the lending rates would go down reducing the margins, but this is not the case. This could imply that treasury bill rates offered by the government are higher and more attractive than the lending rates. Therefore, the banks obtain higher returns from investing in Treasury bills than giving out loans.
234
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
The bank size which in this study indicates market power is found to be positive for Kenya, Japan, Egypt and Botswana with the coefficients all statistically different from zero. This implies that bank concentration have significantly widened the interest margins in these countries. The other countries including USA, Korea, S. Africa, Uganda, Poland, Germany and Malaysia all have negative coefficients. This fact may be indicative of lowering of margins as bank concentration increases probably due to increased efficiency and economies of scale. However, the coefficients are not significant for Malaysia and Poland implying that bank size is not important in explaining the margins in those countries. The coefficients for Uganda, South Africa and USA are all significant at 10% significant level. Among the countries with positive coefficients, the magnitude of the coefficients is highest in Kenya; higher than both Egypt and Japan. This implies that countries with a more concentrated market structure also experience higher interest margins. Market concentration is therefore one of the factors that explain why Kenya’s margins are higher (financial sector is more inefficient) than some of the other countries in the study. Botswana on the other hand has a bank size coefficient that is higher than Kenya’s implying that Botswana’s financial market is more concentrated than Kenya’s. This result is not surprising since Botswana’s financial sector is more inefficient than Kenya’s (wider spreads than Kenya). The cost inefficiency in the banking industry reflects the degree of inefficiency in delivering banking services. The coefficients for all the countries are positive indicating that increased cost inefficiency widens the margins in the respective countries. Cost inefficiency is important in explaining interest rate margins in Kenya, USA, Korea, Germany and Japan with the coefficient for Poland significant at 10% level. Compared to these other countries Kenya has the highest cost inefficiency impact on the margins at 69%. It is not surprising that these countries also have lower spreads than Kenya. This implies that if a country is less cost-inefficient, it will experience wider margins. Inefficiency in delivering financial services therefore is another of the major reasons why Kenya’s financial sector is more inefficient compared to the other countries. The results further show that increased activity and returns from the stock markets improves the efficiency (reduces inefficiency) of the financial markets in all the countries under study as it increases competition for deposits with the banks. However, the coefficient is not significant for Egypt, Malaysia and Botswana. But since the impact of the stock market activity in reducing financial market inefficiency is lower in the USA for instance than in Kenya, while the spreads for USA are at the same time lower than Kenya’s, it cannot be concluded that a more active stock market is the reason why USA’s margins are narrower. If that was the case, the reduction in inefficiency due to the improved activity in the stock market would be higher in the USA than Kenya. Therefore, stock market activity is an important factor in reducing financial sector inefficiency, but is not one of the reasons why the financial markets in Kenya are more inefficient than the other countries. Except for Kenya and Egypt, which have statistically insignificant coefficients for treasury bill rates, all the other countries used in the study indicate that treasury bill rates, which are an alternative investment window for public funds are important in explaining the bank interest rate margins in the respective countries. The coefficients of Japan, Germany, Malaysia and Poland are significant at the 10% level. In general, the results show that in terms of magnitude, cost inefficiency is the major contributor to financial market inefficiency in Kenya followed by financial market structure, and default risk in that order. Stock market activity is an important factor in reducing financial sector inefficiency in Kenya, but is not one of the reasons why the financial markets in Kenya are more inefficient
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
235
than the other countries. Fiscal pressures on the other hand are found to be insignificant in driving financial market in/efficiency. The beauty of assuming different coefficients in the panel as can be seen from the above results is that the factors that contribute to market inefficiency in one country are not necessarily the same as the factors that lead to market inefficiency in the other countries and the magnitude of the impact of a variable differs from one country to the other. 5. Conclusions Widening interest rate margins, which is the indicator of financial sector inefficiency in this study is hypothesized to be detrimental for savings mobilization and stifles investment growth. Wide interest margins as witnessed in Kenya are therefore a sign of a repressed and inefficient financial sector. This paper uses panel cointegration and data from 1990 to 2007 to analyze the determinants of the efficiency of the financial sector intermediation process in 11 countries with a view to recommending policy options for reducing the spreads and improving the financial market efficiency in Kenya. The countries are selected from the African region and other developed countries and includes Kenya, South Africa, Uganda, Egypt, Japan, Germany, Malaysia, Poland, Korea, Botswana and USA. The results show that cost inefficiency is important in explaining interest rate margins in Kenya, USA, Korea, Germany and Japan and Poland with the highest impact of cost inefficiency recorded in Kenya at 69%. It is important to note that these other countries have lower spreads than Kenya. It can therefore be inferred from this result that the other countries have lower margins than Kenya because the impact of cost inefficiency on their margins is not as huge as Kenya’s. In terms of the magnitude, cost inefficiency is found to be the major determinant of wide margins (market inefficiency) in Kenya. The results also show that the impact of default risk is positive and statistically significant for Kenya. The only countries where default risk is not a problem are South Africa and Poland. Important is the finding that the magnitude of impact of default risk on interest spreads for Kenya is higher than those of all the other countries except Uganda and Botswana. All the other countries except Uganda and Botswana have lower spreads than Kenya. It can be inferred again from this result therefore that default risk is one of the major contributors to the financial market inefficiency in Kenya compared to the other countries. This is because the magnitude of impact is higher in Kenya and therefore margins are higher in Kenya compared to the other countries. Default risk is found to be the second most major contributor to inefficient financial markets in Kenya. The third major factor that is important in explaining the financial market inefficiency in Kenya is bank size (financial market structure). Among the countries with positive coefficients, Kenya reports the second highest magnitude of the impact of financial market structure on market inefficiency. Botswana has the highest coefficient on bank size among the countries with positive coefficients but Botswana also has higher inefficiency (wider spreads) than Kenya. This implies that countries with higher bank size (a less competitive market) would experience wider interest margins than countries with more competitive structures. The result here implies that the nature of the financial market structure in Kenya is one of the major factors that explains the very high interest margins (inefficient financial sector) in Kenya compared to the other countries in the study except Botswana. Stock market activity is found to be an important factor in reducing financial sector inefficiency in Kenya but does not account for cross-country differences in market inefficiency between Kenya
236
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
and the other countries. Fiscal pressures on the other hand are found to be insignificant in the Kenyan panel and therefore do not explain the differences in market efficiency between Kenya and the other countries. The results show that the factors that lead to market inefficiency in one country are not necessarily the same as the factors that lead to market inefficiency in the other countries. 5.1. Policy recommendations The findings show that one the main reason why interest rate spreads are high in Kenya compared to the other countries are high cost inefficiencies. To reduce bank operating costs, the banks need to put in place technologically innovative ways of doing business. These include the use of phone banking and e-banking. The recent initiatives towards automation of bank services are a step towards in the right direction, but more still needs to be done. For instance, most teller machines still cannot be used to make deposits but only withdrawals. The second major reason why financial markets in Kenya are more inefficient compared to other countries is high default risk. To make the default risk in the credit market low, banks and the government through the central bank must establish a credit reference system to identify customers who are not credit worthy. There is a credit reference bureau in Kenya, but it lacks the regulatory environment and the support of both the government and the banking sector. The results further show that financial market structure is one of the major contributors to inefficiency in the financial sector in Kenya. The CBK has closed doors to licensing of new banks over the last few years. In this period only K-Rep Bank and Equity bank has been licensed to operate as mainstream banks. Closing the door to new entrants is a sure way of blocking competition from new entrants. The government through the CBK must therefore come in to license more new banks for increased competition. Appendix A. See Tables A.1 and A.2.
Table A.1 Summary of panel unit root tests. Variable
Margins Concentration Cost inefficiency Bank size Bank liabilities Bank liquidity Default risk Share turnover Tbill rates a
Null: unit root
Conclusion
ADF
PP
−0.37 −4.00a 0.26 0.21 1.31 1.03 2.72 4.64 −0.08
1.17 −2.21a −4.65 0.61 0.79 2.50 2.99 10.13 −0.92
Rejection of null hypothesis (of non-stationarity) at 5% with a test statistic of −1.645.
Non-stationary Stationary Non-stationary Non-stationary Non-stationary Non-stationary Non-stationary Non-stationary Non-stationary
Table A.2 SUR results.
Japan Germany Egypt Malaysia Poland Uganda South Africa Korea Botswana USA
Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic Coefficient t-Statistic
Bank liquidity
Bank liability
Default risk
Cost inefficiency
Share turmover
Treasury bills
0.61 7.25 0.09 16.55 −0.01 −2.55 0.03 1.92 −0.01 −0.80 −0.01 −0.38 −1.51 −1.42 −0.01 −1.34 −0.12 −3.99 2.78 6.63 −0.02 −1.20
0.09 1.75 −0.03 −4.54 −0.004 −1.34 −0.19 −6.36 −0.07 −1.91 −0.65 −2.79 −1.74 −1.30 0.02 0.64 0.05 6.81 0.30 2.47 0.43 2.78
0.31 0.08 0.001 0.25 0.07 3.13 0.13 5.71 0.02 1.85 0.52 1.71 2.85 2.15 0.68 2.67 0.02 5.39 0.23 3.37 0.45 2.82
0.64 10.36 0.11 8.02 0.04 2.47 0.04 4.19 0.05 3.69 0.08 0.06 4.27 1.69 0.01 0.07 0.07 2.18 3.31 7.05 0.03 3.17
0.69 2.10 0.18 2.98 0.35 3.79 0.38 0.98 0.03 0.54 0.24 1.36 0.03 0.16 0.12 0.39 0.12 5.30 0.19 0.60 0.28 2.08
−0.19 −2.22 −0.003 −2.14 −0.001 −1.26 −0.0003 −0.10 −0.001 −0.32 −0.01 −1.48 −4825.51 −1.15 −0.17 −3.14 −0.002 −7.13 −0.01 −0.48 −0.002 −1.92
−0.0004 −0.06 −0.02 −1.15 −0.03 −1.38 −0.01 −0.68 −0.11 −1.44 −0.08 −1.24 −0.11 −2.35 −0.68 −2.99 −0.06 −7.94 −0.42 −5.25 −0.12 −3.87
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
Kenya
Banking size
237
238
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
Appendix B. See Figs. B.1–B.6. 40 35
percent
30 25 20 15 10
2007
2008
2006
2004
2005
2002
2003
2001
2000
1999
1997
1998
1996
1995
1993
1994
1992
1991
0
1990
5
Year Deposit rate
Lending rates
Source: International Financial statistics (IFS)-February 2008
Fig. B.1. Interest rate Margins in Kenya (1990–2007). International Financial statistics (IFS) – February 2008. 25
percent
20 15 10
2006
2004
2005
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
0
1990
5
Year Kenya
Korea
Singapore
South Africa
United States
Source: IFS-February 2008
Fig. B.2. Interest rate margins in Kenya and some selected countries of the world. IFS – February 2008. 30
percent
25 20 15 10 5 2006
2005
2004
2003
2002
2001
2000
1999
1998
1997
1996
1995
1994
1993
1992
1991
1990
0
Year Kenya
Egypt
South Africa
Tanzania
Uganda
Source IFS- February 2008
Fig. B.3. Interest margins between Kenya and other regional countries. IFS – February 2008.
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240 Guardian Bank 1% NIC bank 1%
AEBS Bank 3%
Standard Chatered Bank 3%
HFCK 4%
Others 13%
Barclays Bank of Kenya 9% Kenya Commercial Bank 12%
National Bank of Kenya 36%
Co-operative Bank of Kenya 18%
Source: Banking Survey (2007)
Fig. B.4. Non-performing loans in Kenya. Banking Survey (2007).
Fig. B.5. Market share of Kenyan Banks. Banking Survey (2007, pp. 146). Nedbank 9%
Other 5%
Standard Nabk 30%
FNB 20%
ABSA 36%
Source: SACC (2007) Fig. B.6. South African Banks Market Shares (June 2007). SACC (2007).
239
240
J. Oduor et al. / Journal of Policy Modeling 33 (2011) 226–240
References Afanasieff, T. S., Lhacer, P. M., & Nakane, M. I. (2001). The determinants of bank interest spread in Brazil. In Proceedings of the XXIX Encontro Nacional da Economia, Salvador ANPEC. Angabazo, L. (1997). Commercial bank net interest margins, default risk, interest-rate risk, and off-balance sheet banking. Journal of Banking and Finance, 21, 55–87. Barajas, A, Steiner, R., & Salazar, N. (1999). Interest spreads in banking in Colombia. IMF Staff Papers, 46(2), 196–225. CCSA. (2008). Banking Industry Survey. Pretoria, South Africa: CCSA. Demirguc-Kunt, A., Laeven, L., & Levine, R. (2003) The impact of bank regulations, concentration, and institutions on bank margins. Policy Research Working Paper Series 3030, The World Bank. Demirguc-Kunt, & Hiuzinga, H. (1997). Determinants of commercial bank interest margins and profitability: Some international evidence. World Bank working paper No. 1900, Washington DC. Granger, C. W. J., & Newbolt, P. (1974). Spurious regressions in econometrics. Journal of Econometrics, 2, 111–120. Im, K. S., Pesaran, M. H., & Shin, Y. (1995). Testing for unit roots in panel data. Department of Applied Economics, University of Cambridge. Manuscript. Kenya Bankers Association (KBA) (2000). A statement by KBA to the Parliamentary Finance Committee, on the Central Bank (Amendment) Bill, 2000. Ndung’u, S. N., & Ngugi, R. W. (2000). Banking sector interest rate spread in Kenya. KIPPRA paper DP/05/2000. Zellner, A. (1962). An efficient method of estimating seemingly unrelated regression equations and tests for aggregation bias. Journal of the American Statistical Association, 57, 348–368.