台灣銀行業國際化與風險之關係探討 - 政大學術集成

全文

(1)國立政治大學商學院金融學系研究所碩士學位論文. 台灣銀行業國際化與風險之關係探討政治. 大. 立. ‧ 國. 學. Risk of Internationalization on Taiwanese Banking Industry. ‧. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. 指導教授 : 李桐豪博士研究生 : 鄭宇淳撰. 中華民國 104 年 6 月.

(2) 謝辭. 感謝李桐豪老師耐心指導與我討論研究，以至完成本篇碩士論文，吾實獲益良多。在每次的討論中，老師嚴謹的思考邏輯以及豐富的實務經驗，為我提供更精準的研究方向，並且解答我研究中的許多疑惑，豐富我兩年的碩士生涯。除學術上的討論，在老師身上也學習到許多生活哲學以及人生素養，是我在未來人生道路上奉行的圭臬。感謝口試委員張興華老師、馮立功教授給予建設性的意見，使我整篇論文更. 政治大. 臻完善；衷心感謝遲淑華學姊，撰寫論文的一路上全力支持與協助，並且多次的. 立. 細心叮嚀，只為能讓我順利通過口試，完成論文，遲學姊無私的貢獻與分享，吾. ‧ 國. 學. 莫生難忘；感謝同學們、朋友們的寬容量解，在兩年研究所的時光給予我鼓勵與動力，以至完成學業。. ‧. 最後，感謝父母和長兄的無私奉獻，讓我無憂的放手追求知識，體驗人生。. y. Nat. sit. 雖然一人在北部求學，加上工作的緣故，常常久日無法相見，但家人的心還是緊. n. al. er. io. 繫在一起，彼此互相扶持。求學之路暫時在此畫上句點，未來的日子希望能依舊. i n U. v. 抱持著樂觀的心、堅定的毅力與冒險的勇氣，勇闖人生。. Ch. engchi.

(3) 摘要. 隨著全球金融市場的開放，多數金融機構開始向海外擴展業務，追求更高利潤並且取得市場領先的地位。在台灣，銀行業者由於市場規模小、競爭激烈加上存、放款利差逐漸縮小的情況下，許多銀行開始往亞洲其他國家擴展，尤其是中國市場與東南亞市場。許多研究顯示，銀行國際化程度與風險呈現正相關，顯示在銀行向海外布局同時將使其風險增加。台灣過去文獻多在研究銀行業國際化程度與經營績效的關係，對於國際化程度與風險的影響較少著墨，本文使用台灣商. 政治大. 業銀行之資料，進一步研究兩者之關聯。本研究實證顯示，整體而言，台灣銀行. 立. 業可透過海外布局達到降低風險的效果，但進一步發現，台灣商業銀行資產規模. ‧ 國. 學. 大小將會造成國際化與風險的關係有所不同。大型商業銀行，雖然整體風險較小，但當國際化程度越高時，將使得風險提高；反之，小型商業銀行，整體風險較大，. ‧. 但可透過海外布局使其風險下降。. n. er. io. sit. y. Nat. al. i n U. v. 關鍵字: 台灣、商業銀行、銀行國際化、風險、資產規模. Ch. engchi.

(4) Abstract. Global banking system has changed dramatically in the last several decades which make banks encounter difficulties and discover opportunities while developing their business. In this increasingly saturated environment, existing banks start to find a way to expand their territories overseas to set up higher entry barriers. Especially in Taiwan, because of excessive banks, small economic scale and overlap financial services, the financial industry is more competitive than in other markets resulting the interest rate. 政治大. spread becomes more and more narrow. Most of research find a positive relation. 立. between internationalization and bank risk. However, this paper employs the data of. ‧ 國. 學. commercial banks in Taiwan, the result suggests that internationalization of a bank is associated with lower risk for overall banks. Additionally, we find that the bank size. ‧. has different impacts on the relationship between internationalization and risk. Large. y. Nat. io. sit. bank tend to have the lower average risk but to increase the bank risk when go abroad.. n. al. er. On the other hand, small banks have a diversification effect on internationalization but. i n U. v. the overall risk is larger and the operation is not as efficient as large banks.. Ch. engchi. Keywords: Taiwan, Internationalization, Risk, Commercial Bank, Size.

(5) Content Chapter 1 Introduction………………………………………………………………...1 Chapter 2 Literature Review…………………………………………………………..3 2.1 Internationalization of Bank………………………………………………….3 2.2 Bank Risk…………………………………………………………………….4 Chapter 3 Methodology and Framework…………………………………………........5 3.1 Data………………………………………………………………………….. 5 3.2 Measure of Risk……………………………………………………………5. 政治大 3.4 Control Variable……………………………………………………………...7 立 3.3 Measure of Internationalization………………………………………………6. ‧ 國. 學. 3.5 Empirical Model…………………………………………………….............10 Chapter 4 Empirical Analysis………………………………………………………...12. ‧. 4.1 Summary Statistics………………………………………………………….12. sit. y. Nat. 4.2 Univariate Analysis…………………………………………………………15. al. er. io. 4.3 Regression Analysis………………………………………………………...16. v. n. 4.4 Robustness Tests…………………………………………………………….19. Ch. engchi. i n U. 4.4.1 Alternative Measures of Risk………………………………………..19 4.4.2 Alternative Measures of Internationalization………………………21 4.5 Additional Analysis…………………………………………………………22 4.5.1 Internationalization and Risk during Financial Crisis……………….22 4.5.2 Bank Size Break Down……………………………………………23 Chapter 5 Conclusions……………………………………………………………….26 Reference……………………………………………………………………………27. I.

(6) List of Tables Table 1 Financial Industry Contribution of GDP in Taiwan…………………………...1 Table 2 The Banks in Taiwan………………………………………………………….5 Table 3 Variable Definitions…………………………………………………………...9 Table 4 Summary Statistics…………………………………………………………..14 Table 5 Correlation Analysis…………………………………………………………15 Table 6 Internationalization and Bank Risk: Univariate Analysis…………………...16 Table 7 Internationalization and Bank Risk: Regression Analysis…………………...18. 政治大 Table 9 Alternative Measures 立of Internationalization………………………………..21 Table 8 Alternative Measures of Risk………………………………………………19. ‧ 國. 學. Table 10 Internationalization and Bank Risk during Financial Crisis………………22 Table 11 Regression Analysis by Different Bank Size……………………………….24. ‧. Table 12 Z-score Decomposition by Different Bank Size…………………………….25. List of Figures. er. io. sit. y. Nat. al. n. v i n Ch Figure 1 Distribution of Z-score……………………………………………………...12 engchi U. Figure 2 Distribution of Logarithm of Z-score……………………………………....13 Figure 3 Distribution of Logarothm of Total Asset…..................................................24. II.

(7) 1 Introduction Global banking system has changed dramatically in the last several decades which make banks encounter difficulties and discover opportunities while developing their business. In this increasingly saturated environment, existing banks start to find a way to expand their territories overseas to set up higher entry barriers. In Taiwan, because of excessive banks, small economic scale and overlap financial services, the financial industry is more competitive than in other markets resulting the interest rate spread becomes more and more narrow. Consequently, internationalization. 政治大 1, after the financial crisis 立 the weighting of Taiwanese GDP in financial industry has. is the one of main solutions for bank to gain more profits in Taiwan. As shown in Table. ‧ 國. 學. dramatically declined from 7.01% in 2008 to 6.19% in 2010. However, the weighting is increasing gradually in recent years due to Taiwanese government has aggressively. ‧. advocated the development of financial industry. In 2009, Taiwan signed the. sit. y. Nat. Memorandum of Understanding on Cross Strait Banking (MOU) with China, allowing. n. al. er. io. its banks to tap China's massive market and paving the way for banks on both sides to. i n U. v. invest in each other. As a result, Taiwanese banks started to set up representative offices. Ch. engchi. in the first-tier cities of China. In the following year, Taiwan signed the Economic Cooperation Framework Agreement (ECFA) with China in Chongqing city to lower tariffs and to relax access for 539 products and services for Taiwan. Obviously, these two policies have a significant impact on financial industry in Taiwan. Table 1 Financial Industry Contribution of GDP in Taiwan NTD (1 billion). 2008. 2009. 2010. 2011. 2012. 2013. 2014. GDP - Total GDP - Financial Industry Financial Industry/GDP. 13,151 913 7.01%. 12,962 798 6.23%. 14,119 872 6.19%. 14,312 915 6.39%. 14,687 933 6.42%. 15,221 970 6.45%. 16,084 1,042 6.61%. Source: Directorate General of Budget, Accounting and Statistics, Executive Yuan, R.O.C.. 1.

(8) However, while banks seek more profit overseas, do banks diversify their asset allocation to lower the risk or to take on more risk from market specific factor? Some evidences show that international banks can diversify their risk through assessing to global financial market which supports the diversification hypothesis (Doukas and Kan, 2006; Laeven and Levine, 2007). Alternatively, market risk hypothesis suggests that internationalization increases banks’ risk (Amihud, Y., DeLong, G. and Saunders, A., 2002). Further, Berger, A., Sadok El Ghoul, Omrane Guedhami and Raluca A. Roman (2013) use the data among the U.S. banks from 1989:Q1 to 2010:Q4 and test with. 政治大 the bank, the higher the risk. Our paper follows Berger, A., Sadok El Ghoul, Omrane 立 varieties of robustness models which strongly suggest that the more internationalized. Guedhami and Raluca A. Roman (2013) to construct the research structure to. ‧ 國. 學. investigate the relationship between internationalization and risk of banks in Taiwan. ‧. using a sample of 1,991 bank-quarter observations from 2004:Q4 to 2014Q2. We find. sit. y. Nat. that internationalization of a bank is associated with lower risk for overall Taiwanese. io. er. banks. But when we segment our sample into large banks and small banks, we find that in each subsample the relations between internationalization and banks’ risk are. al. n. v i n different. In large banks, which C have a lower risk profile h e n g c h i U tend to take on more risk with higher degree of internationalization; in small banks, which have a higher risk profile are benefited from asset diversification abroad. The paper proceeds as follows. Section 2 reviews some research on internationalization and risk. Section 3 describes the data, variables and empirical methodology. Section 4 presents summary statistics, regression result, robustness test and some additional analysis. Section 5 is conclusion.. 2.

(9) 2 Literature Review Internationalization does not mean to give up domestic market, but to provide multinational corporations with banking services so that banks can hold their national position and can have more capable of ensuring banking expansion. However, how bank internationalization activity affects the risk has been wildly concerned. 2.1 Internationalization of bank According to Annavarjula and Beldona (2000), there are three dimensions defining internationalization: Operation, Ownership and Orientation. The operation. 政治大 on foreign market, such as 立a number of foreign subsidies of total subsidies. The. dimension is reflecting the foreign market penetration of a firm and its independence. ‧ 國. 學. ownership dimension is reflecting the foreign production of a firm. The orientation dimension represents the business strategy, structure and management style are. ‧. international-oriented. For banks, there are two purposes to go internationalization, one. sit. y. Nat. is international participation and another one is international expansion. Aliber (1976). al. er. io. indicates when a bank has more comparative advantage on financial services, it is more. v. n. likely developing into an international bank base on ‘International Trade Theory’;. Ch. engchi. i n U. Moreover, domestic market density may also increase a bank’s operation efficiency to take advantage of entering global market. Based on banking activities, three different types of multinational banks (Grubel, 1977) can be identified as a.) Multinational Retail Banking: this type of banks provides individual banking services, such as personal loans, credit cards and mortgage services, in oversea markets. The management expertise and marketing know-how will be the main advantage to gain customers. b.) Multinational Service Banking: giving multinational corporations financial services for supporting their business is substantial. Long-term customer relationship is much more important in this type of banks which 3.

(10) let local banks hard to compete with. c.) Multinational Wholesale Banking: banks help customers to deal with huge cash transactions between borders by Eurocurrency market to minimize cost. This kind of banks has economic of scale on investment information on financial consulting. 2.2 Internationalization and Bank risk Some argue that international banks can diversify their risk through assessing to global financial market (Doukas and Kan, 2006; Laeven and Levine, 2007), and Buch, Claudia M., John C. Driscoll and Charlotte Ostergaard (2010) show that from a. 政治大. mean-variance point of bank estimated gains considerably from allocating bank asset. 立. in to cross-border.. ‧ 國. 學. However, others have another view on the impact of internationalization on risk. International banks can take advantage of portfolio diversification reducing their risk,. ‧. but this effect can be offset by incentives going in the opposite direction leading them. sit. y. Nat. to take on excessive risks (Gulamhussena, M.A., Pinheirob, C., and Pozzolo A.F., 2014).. n. al. er. io. Berger, A., Sadok El Ghoul, Omrane Guedhami, and Raluca A. Roman (2013) also find. i n U. v. that a positive relation between internationalization and bank risk, and the result is. Ch. engchi. robustness examined by different measures and econometric models. To analyze the internationalization of bank risk, we use Z-score (e.g., Boyd and Runkle, 1993; Laeven and Levine, 2009; Berger, 2013) to measure which is defined as the sum of a bank’s mean return on assets and mean on capitalization ratio divided by the standard deviation of return on assets. The Z-score is a commonly method to measure risk and has been used in numerous of the empirical banking literatures to reflect insolvency risk which captures the distance of default. Z-score also assesses both individual bank risk and overall financial stability which can be easily computed using accounting data. Therefore, we also follow this measure to evaluate risk. 4.

(11) 3 Methodology and Framework 3.1 Data We focus on commercial banks in Taiwan, and assemble quarterly data from Central Bank of Republic China (Taiwan) which contains income statements, balance sheets (also refer to statements of financial position), and major financial ratios of all Taiwanese commercial banks. Data covers the period from 2004:Q4 to 2014:Q2 with 1,991 bank-quarter observations for 79 banks over the entire sample period. According to the report from Central Bank of Republic China (Taiwan), there are 71 commercial. 政治大 subsidiaries in 2014. Table立 2 is the break down summary of banks in Taiwan.. 學. Taiwan Taiwan Taiwan. io. al. n. Foreign Bank China Bank Holdings. Nat. Domestic Bank Taiwan Domestic Bank Oversea. Head Office. OBU. Branch. 41 -. 38 0. 3,451 124. 27 3 16. 24 0 0. 36 3 0. Ch. engchi. Representative Subsidiary Office. y. Region. er. Financial Company. sit. Table 2 Financial Statistics in Taiwan. ‧. ‧ 國. banks, 16 financial holdings, 62 OBU, 3,614 branches, 50 representative offices and 20. i n U. v. 3 47. 0 20. 0 0 0. 0 0 0. Source: Banking Bureau, Financial Supervisory Commission, ROC. 3.2 Measures of risk Follow by Berger, Allen N., Sadok El Ghoul, Omrane Guedhami, and Raluca A. Roman (2013), our primary measure of bank risk-taking is Z-score assessed as the average return on assets plus the average capitalization asset ratio (Equity/Total asset), divided by the standard deviation of return on asset. Standard deviation of return on assets is calculated over the previous 12 periods. Z-score has been widely used in the recent literature (Boyd and Runkle, 1993; Boyd, De Nicolo, and Jalal, 2006; Beltratti and Stulz, 2010). However, some researches declare that Z-score is highly skewed, we 5.

(12) take the natural logarithm of the Z-score (Laeven and Levine, 2009; Houston et al, 2010; Bhagat S., Bolton, B. and Lu J., 2012) to cope with this concern. We also construct Zscore over previous 8 quarters and 20 quarters to use in robustness test. In addition, several standard measures of risk based on accounting data are used in our study. 1) Standard deviation of return on equity (Stdv. ROE): This is evaluated over 12 quarters, where return on equity (ROE) is net income divided by total equity. 2) Risk-adjusted rate of return on equity (RAROE): RAROE is defined as the ratio of the average ROE to its standard deviation by using rolling basis of 12 quarters. 政治大 Also, two variable are used to measure stability and sustainable growth of banks (Ana 立 (Chiorrazo et al. 2008; Demirgüç-Kunt and Huizinga 2010).. Rosa Fonseca and Francisco González 2010; Berger, Allen N., Sadok El Ghoul,. ‧ 國. 學. Omrane Guedhami, and Raluca A. Roman 2013). 1) Non-performing loans ratio. ‧. (NPLR): A loan is non-performing when payments of interest and principal are past due. sit. y. Nat. for over 90 days; NPL ratio is computed as the ratio of non-performing loans to total. io. er. loans. Flamini (2009) finds that the success of individual banks in credit risk management is largely reflected in the proportion of NPL’s loans to gross lending. 2). al. n. v i n C h The ratio of allowance Allowance for loan loss ratio (LLAR): for loan losses to total engchi U loans.. 3.3 Measures of internationalization Following Cetorelli and Goldberg (2012) and Berger, Allen N., Sadok El Ghoul, Omrane Guedhami, and Raluca A. Roman (2013), we use several methods to evaluate the degree of bank internationalization. 1) Foreign Assets Ratios (FAR): The main measure of bank internationalization computed as a bank’s foreign asset divided by total asset. 2) Bank Internationalization Dummy: If a bank has assets oversea then we take the value 1, 0 otherwise. 3) Foreign Loans Ratio (FLR): The ratio of a bank’s foreign 6.

(13) loans to the total loans of the bank. 4) Foreign Deposits Ratio (FDR): A bank’s foreign deposits to the total deposits of the bank. This study uses foreign assets ratio as the primary measurement for internationalization, and the others are used in robustness tests. 3.4 Control variable Many of prior researches identify that bank characteristics may affect the bank risk exposure and might potentially bias our results. Therefore, in order to account for these factors, we include several time varying bank-specific controls in our investigation.. 政治大 portfolio diversification, but 立this effect can lead them to take on more excessive risk.. 1) Income Diversification (DIV): Internationalization bank can benefit from. ‧ 國. 學. Thi Canh Nguyen, Dinh Vinh Vo, and Van Chien Nguyen (2015) find that a bank with high non-interest income presents lower risk than those with mainly interest income.. ‧. Demirgüç-Kunt and Huizinga (2010) and Baele, De Jonghe, and Vander Vennet (2007). sit. y. Nat. find that a higher reliance on non-interest income is linked to more risk adjust returns.. al. n. Rumble (2006).. er. io. To evaluate the diversification level, we follow the approach of K. J. Stiroh and A.. Ch. engchi. i n U. v. Commercial banks’ activities are classified between traditional (taking deposits and making loans) and non-traditional (e.g., security and foreign exchange trading and provision of fee-based services). We consider income diversification into two categories, net interest income (NII) and non-interest income (NOI). Net interest income is interest income minus interest expense and non-interest income including investment income, foreign exchange income, gain (or loss) on sale of securities, trading account income and commissions and fees. Afterward, Herfindant-Hirschman Index (HHI) is used to measure the diversification level which is highly practical and is used widely to measure the competitiveness in a specific industry or market. We 7.

(14) NII. 𝑁𝑂𝐼. construct income diversification as1 − [( )2 + (𝑁𝐼𝐼+𝑁𝑂𝐼 )2 ]. The smaller the 𝑁𝐼𝐼+𝑁𝑂𝐼 value is, the lower diversification level is and vice versa. 2) Size: The log of total assets as measure of the bank size. Evidences show that bigger banks tend to better survive in competition to have higher asset-quality and to be more leveraged drives banks into a race for expansion (Tianxi Wang, 2014). Futhermore, the large banks are much more diversified and stable taking advantages on economics of scale (Demsetz, R. S. and P. E. Strahan, 1997; Berger, Bouwman, Kick, and Schaeck, 2012). However, large banks are riskier, and create more systemic risk. 政治大 in market-based activities or 立are more organizationally complex (Laeven L., Ratnovski. ethier when they have lower capital and less-stable funding or when they engage more. ‧ 國. 學. L., and Tong H., 2014). Mamiza Haq and Richard Heaney (2012) also show that large banks may reflect higher total risk and lower credit risk.. ‧. 3) Listed: This study measures listed as a dummy variable which equal to one if. sit. y. Nat. the bank is listed in Taiwan Stock Exchange, 0 otherwise. Listed banks are considered. n. al. er. io. by rating agencies to have a higher default risk (Giuliano Iannottaa, Giacomo Nocerab, and Andrea Sironi 2013).. Ch. engchi. i n U. v. 4) Financial Holding Company (FHC): We set up a variable to control the outcome of bank risk in our study, taking value of 1 if the bank is part of a financial holding company, and 0 otherwise. In November 2001, the ‘Financial Holding Company Act’ was implemented in Taiwan and that provides financial companies to function as a management umbrella by operating cross-industry, such as banking, securities and insurance. As of 2014, the government approved the application of 16 financial institutions to set up financial holding companies. Laeven and Levine (2007) show that the diversification discount exists in a financial conglomerate instead of diversification premium. Stiroh and Rumble (2006) 8.

(15) demonstrate that non-interest income has a negative impact on banks’ diversification in the U.S. financial holding companies. In Taiwan, an independent bank could have lower systemic risk because an FHC bank with diversified activities will reinforce the agency problem between insiders and outsiders (Shen, C. H. and Chang Y., 2012). 5) Overhead Cost (OHC): In order to capture the bank’s operating cost structure, our study evaluates it as the ratio of total bank operating expenses to total asset named overhead cost. Demirgüç-Kunt, A. and Huizinga, H. (2010) show that banks with high overheads are further estimated to be less stable and fee-generating activities are. 政治大 Table 3 lists the definitions of all variables that are used in our research. 立. relatively costly.. ‧ 國. 學. Table 3 Variable Definitions. ‧. Definition. Nat. n. al. Stdv. ROE. RAROE. NPL Ratio (NPLR) LLA Ratio (LLAR). sit. Risk Variables. A bank measure of financial risk calculated as logarithm of [Avg.(ROA)+Avg.(Equity/Total Asset)]/Stdv.ROA; a larger value indicates lower overall bank risk. Means of ROA and Equity/GTA as well as the standard deviation of ROA are computed. er. io Z-score. y. Variable. Ch. engchi. i n U. v. over the previous 12 quarters (t-11 to t). The standard deviation of ROE calculated over the previous 12 quarters (t-11 to t.) ROE is determined as the ratio of operating income to total equity. The risk-adjusted return on equity defined as ROE/Stdv.ROE. ROE is determined as the ratio of net operating income to total equity by using rolling basis of 12 quarters (t-11 to t.). The ratio of non-performing loans (past due at least 90 days or in nonaccrual status) to total loans. The ratio of allowance for loan losses to total loans. 9.

(16) Internationalization Variables A measure of bank internationalization determined as the ratio of foreign total assets to total assets of the. Foreign Asser Ratio (FAR). Bank Internationalization Dummy. Foreign Loans Ratio (FLR). bank. A dummy variable that takes a value of 1 if a bank exists foreign assets, and 0 otherwise. A measure of bank internationalization determined as the ratio of foreign total loans to total loans of the bank. A measure of bank internationalization determined as the ratio of foreign total deposits to total deposits of the bank.. Foreign Deposits Ratio (FDR). 治 of diversification across different sources 政A measure 大. Main Bank Characteristics. 立. of. income,. calculated. ‧ 國. 𝑁𝑂𝐼. (𝑁𝐼𝐼+𝑁𝑂𝐼 )2 ]; where NII is net interest income and. ‧. NOI is non-interest income. A measure of bank size determined as the log of total assets.. Size. y. Nat. sit. A dummy variable that takes a value of 1 if the bank is listed on a stock exchange or is part of a bank holding company that is listed on a stock exchange and 0 otherwise. A dummy variable that takes a value of 1 if the bank. n. al. er. io. Listed. 學. Income Diversification (DIV). NII. as 1 − [(𝑁𝐼𝐼+𝑁𝑂𝐼 )2 +. Ch. Financial Holding Company (FHC) Overhead Cost (OHC). engchi. i n U. v. is owned by a bank holding company, and 0 otherwise. A proxy for the bank's cost structure calculated as the ratio of overhead expenses to GTA.. 3.5 Empirical model To investigate the data feature, we employee univariate analysis. We compute mean, medium, standard deviation, and two samples t-test to derive the descriptive statistics In order to prevent from collinearity in further analysis, correlation analysis is used. 10.

(17) Next, we conduct Ordinal Least Square (OLS) multivariate regression model to estimate the relationship between internationalization of bank and risk. Meanwhile, in order to control the individual bank’s characteristics and the seasonal effect, we include important control variables and seasonal dummy variables. We use following model to estimate our study: Risk 𝑡−𝑘+1,𝑡 = 𝛼 + 𝛽1 × 𝐼𝑛𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖𝑧𝑎𝑡𝑖𝑜𝑛𝑡−𝑘 + 𝛽2 × 𝐵𝑎𝑛𝑘 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠𝑡−𝑘 + 𝑇𝑖𝑚𝑒 𝐹𝑖𝑥𝑒𝑑𝑡 + 𝜀𝑡−𝑘+1,𝑡 where we measure the Risk variable over the k quarter from t − k + 1 to t, and. 政治大 predetermined associated with the dependent variable. 立. the independent variables are measured in the quarter t − k to ensure that they are. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 11. i n U. v.

(18) 4 Empirical Analysis In this session, we empirically analyze the impact of internationalization on bank risk. First, we simply descript the summary statistics of our variables and some characteristics of the banks. Second, univariate tests are used to compare the difference of international bank and domestic bank. Last, we conduct the regression analysis to demonstrate the relationship of internationalization and bank risk under some bank’s control variables. 4.1 Summary statistics. 政治大 and Figure 2 shows that the distribution of logarithm of Z-score. As noted, the 立 Figure 1 shows that the distribution of Z-score without taking the natural logarithm. distribution of Z-score has a serious skewed problem (Laeven and Levine, 2009;. ‧ 國. 學. Houston et al, 2010; Bhagat S., Bolton, B. and Lu J., 2012), and in our sample which is. ‧. skewed to right; therefore, after taking the natural logarithm of Z-score as shown in. sit. y. Nat. Figure 2, the distribution is much more similar with normal distribution. As a result, we. io. al. n. 250. 200. Frequency. er. will use the logarithm of Z-score for our further analysis.. Ch. engchi. i n U. v. 150. 100. 50. 0 8.20. 22.59. 36.98. 51.36. 65.75. 80.13. 94.52. Z-score. Figure 1 Distribution of Z-score 12. 108.91 123.29 137.68.

(19) 300. 250. Frequency. 200. 150. 100. 50. 0 0.35. 1.04. 1.74. 2.43. 3.12. 3.81. 4.50. 5.19. 5.88. 6.57. 政治大 Figure立 2 Distribution of Logarithm of Z-score Logarithm of Z-score. ‧ 國. 學. Table 4 demonstrates the summary statistics among all the variables in our study. In our main measure of bank risk, Z-score, the mean Z-score (12 quarters) is 3.265. It. ‧. indicates that most of commercial banks in Taiwan with an average RAROE of 1.642,. y. Nat. io. sit. an average Stdv. ROE of 0.106, an average NPL Ratio of 3.1% and an average LLA. n. al. er. Ratio of 1.6% have a small chance to default. The NPL Ratio has a very high standard. i n U. v. deviation showing that there is a highly difference of lending situations among banks. Ch. engchi. in Taiwan. The international measure indicates that almost 90% of all the commercial banks in Taiwan, with a mean foreign assets ratio of 9.4%, a mean foreign loans ratio of 8% and a mean foreign deposits ratio of 4.7%, have begun running the business overseas after 2002. In bank characteristics, 46% of the banks are listed in Taiwan Stock Exchange and 27% of the banks are owned by a financial holding company. Also, the average commercial bank has a size of 12.9, an income diversification of -1.165 which is relatively low and overhead costs of 0.008.. 13.

(20) Table 4 Summary Statistics Variable. Mean Median Stdv.. 25p. 75p. 2.872 2.975 2.665 0.621. 3.834 3.880 3.772 2.630. Risk Variables Z-score (12 quarters) Z-score (8 quarters) Z-score (20 quarters) RAROE. 3.265 3.365 3.159 1.642. 3.418 3.473 3.290 1.423. 0.845 0.857 0.812 1.177. Stdv. ROE NPLR LLAR. 0.106 0.031 0.016. 0.034 0.016 0.011. 0.300 0.022 0.062 0.057 0.006 0.029 0.021 0.009 0.015. Internationalization Variables 0.073 0.079. 0.101 0.024 0.130 0.101 0.040 0.136. 0.020. 0.292 1.000 1.000 0.096 0.016 0.101 0.072 0.004 0.060. 政治大 Bank Internationalization Dummy 0.906 1.000 立 FLR 0.080 0.055 Main Bank Characteristics -1.165 12.907 0.458 0.274. 0.358 12.798 0.000 0.000. OHC. 0.008. 0.006. 37.399 1.116 0.498 0.446. 0.264 12.108 0.000 0.000. 0.450 13.905 1.000 1.000. io. 0.007 0.004 0.010. er. Nat. DIV Size Listed FHC. ‧. ‧ 國. 0.047. 學. FDR. y. 0.094 0.104. sit. FAR (Full sample) FAR (International banks only). al. n. v i n C h correlation U Table 5 demonstrates the Pearson e n g c h i for the dependent variable and independent variables which will be used in further empirical analysis. Banks with higher foreign asset ratio tend to have higher Z-score indicating lower risk; large banks tend to have lower risk (higher Z-score) and higher foreign asset ratio. However, a negative relationship has been shown between listed bank and risk where listed banks are more likely to be large size.. 14.

(21) Table 5 Correlation Analysis This table reports correlations for the key bank variables used in the regression analysis. ***, **, and * indicate significance at the 1%, 5%, and 10% levels respectively. Pearson Correlation Coefficient Z-score (12quarters) Z-score. FAR. DIV. Size. Listed. 1 0.0233 -0.0210. 1 0.4134***. 1. FHC. OHC. 1. (12 quarters) FAR DIV Size Listed. 0.24697*** 1 -0.0187 0.0194 0.1409*** 0.2651*** -0.1346*** 0.1439***. FHC OHC. 0.0167 0.1718*** -0.0402* 0.4354*** 0.6090*** 1 -0.3847*** -0.1562*** 0.0205 -0.2649*** -0.0622*** -0.0714***. 立. 政治大. ‧ 國. 學. 4.2 Univariate analysis. In order to clearly investigate the relationship between internationalization and. ‧. risk, we use the bank internationalization dummy to separate our sample into two. sit. y. Nat. groups, international banks and domestic banks. We compare the means of risk. io. er. variables and bank characteristics variables by using two sample t-test. Table 6 indicates. al. v i n 2.963 in domestic banks, and theCdifference h e n giscstatistically h i U significant at the 5% level. n. that international banks have a higher mean Z-score (12 quarters) of 3.288 compared to. The results of using other measures of bank risk consistently suggests that a bank with more overseas operations will take on more risk. For example, the average of Zscore computed over 8 quarters and 20 quarters for international banks are both higher than for domestic banks, and the differences in mean are 0.192 and 0.704, respectively. Furthermore, the RAROE examining the performance by adjusting for the risks is higher for international banks compared to domestic banks with the difference in mean of 0.483. Also, the average of Stdv. ROE is 1.677 for international bank compared to 1.194 for domestic banks. Both the average of non-performance loans ratio and 15. 1.

(22) allowance for loan losses ratio of international banks are lower than domestic banks with the difference in the mean of -3.8% and -0.9%. All of these differences are statistically significant. In sum, our findings support that commercial banks in Taiwan have an advantage on portfolios diversification contributing to lower risk. Table 6 Internationalization and Bank Risk: Univariate Analysis This table reports univariate comparison tests for international banks versus domestic banks. ***, **, and * indicate significance at the 1%, 5%, and 10% levels respectively. International Banks (2). N. Diff.. t-stat. 0.325**. 2.39. 0.192**. 2.44. 129. 3.189. 840. 3.175. 19. 2.471. 0.704***. 3.76. 1,173. 1.677. 90. 1.194. 0.483***. 3.77. Stdv. ROE NPLR LLAR. 1,173 1,756 1,761. 0.092 0.026 0.015. 90 219 225. 0.288 0.065 0.024. -0.196*** -0.038*** -0.009***. -3.42 -6.51 -3.84. DIV Size. 1,761 1,761. Listed FHC OHC. 1,761 1,761 1,761. Nat. io. n. al. er. 3.381. (8 quarters) Z-score (20 quarters) RAROE. ‧ 國. 1,373. y. (6). ‧. (5). sit. 1,173. (3). 學. Z-score (12 quarters) Z-score. International Domestic. (4) 治 Mean政 N Mean 大立3.288 90 2.963. (1) Variable. Domestic Banks. v i n C h 229 -0.375 -1.268 -0.893 U i e h n g c 11.561 1.521*** 13.083 230. -0.85 21.59. 0.509 0.304 0.008. 21.31 14.33 -2.88. 230 230 230. 0.070 0.048 0.010. 0.439*** 0.256*** -0.002***. 4.3 Regression analysis Our goal is to evaluate how internationalization affects the risk taken by banks. We construct several regression models to elaborate our ideas. Table 7 presents the results from regressing Z-score (12 quarters) on internationalization variables with banks’ control variables. 16.

(23) Model 1 includes all the data into regression and reports the results from regressing Z-score (12 quarters) on our main international variable, Foreign Asset Ratio. After controlling the bank characteristic variables and time fixed effects, we find that banks with higher Foreign Asset Ratio tend to have higher Z-score which means that banks can lower the risk through internationalization. An increase in one standard deviation of Foreign Asset Ratio (0.101) is associated with a decrease in Z-score of 0.154. Model 2 and Model 3 present additional results. In Model 2, we replace Foreign Asset Ratio with Bank Internationalization Dummy. In Model 3, we exclude domestic. 政治大 The results show that the coefficient estimated on both internationalization variables 立. banks in our sample to investigate the purely marginal effect of the Foreign Asset Ratio.. are positive and statistically significant at 1% level.. ‧ 國. 學. In bank control variables, we can see that coefficients of Listed and OHC are both. ‧. negative and statistically significant at 1% significance level. However, the coefficient. sit. y. Nat. of Size is not statistically significant. This result shows that the size of banks in Taiwan. io. To sum up, our. al. v i n primaryCstudy suggests thatUbank in hengchi. n. take more research.. er. may not have significant explanatory power toward risk management and should be. Taiwan with higher. internationalization are more capable of lowering the risk. Berger, Allen N., Sadok El Ghoul, Omrane Guedhami, and Raluca A. Roman (2013) using the data of commercial bank in the U.S. find that the market risk hypothesis dominates the diversification hypothesis and banks are more likely to take on more risk when going abroad which is contrary to our studies. That may be due to geographic difference and operations and activities difference, because banks in Taiwan usually follow their clients’ pace and set up businesses in other countries to cope with their needs.. 17.

(24) Table 7 Internationalization and Bank Risk: Regression Analysis This. table. reports. regression. analysis. of. the. relationship. between. the. internationalization of Taiwanese commercial banks and risk. The main measure of internationalization is Foreign Asset Ratio and the dependent variable is our main measure of risk, Z-score (12 quarters). Robust t-statistics are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels respectively. Dependent Variable: Z-score (12 quarters) (1) Full Sample. Independent Variable FAR. (2) Full Sample. (3) International Banks Only. 1.526*** (7.74). International Banks Dummy. 立. 2.928*** (9.81). 3.539*** (12.44). -0.00028 (-0.60) 0.032 (1.42) -0.433***. -0.00018 (-0.37) 0.020 (0.86) -0.464***. -0.00006 (-0.14) 0.006 (0.28) -0.464***. (-8.65) 0.137** (2.47) -76.690*** (-16.54). (-9.05) 0.176*** (3.16) -81.931*** (-17.62). (-9.89) 0.152*** (2.95) -91.330*** (-17.90). Time Fix Effect. Yes. Yes. Yes. Observations R-Square Adj. R-Square. 1,263 0.284 0.278. 1,263 0.266 0.261. 1,173 0.313 0.307. io. OHC. al. n. FHC. Ch. engchi. 18. y. sit. Nat. Listed. ‧. Size. er. DIV. 學. 3.100*** (10.62). ‧ 國. Intercept. 政治 0.534*** 大(5.42). 1.273*** (6.80). i n U. v.

(25) 4.4 Robustness tests 4.4.1 Alternative measures of risk Considering the robustness of our findings, we replace the Z-score with alternative measures of bank risk. First, aside from computing Z-score over previous 12 quarters, we apply 8 quarters and 20 quarters as a dependent variable in Model 2 and Model 3 respectively. Second, in Model 4, we use RAROE to examine bank risk, and in Model 5, Stdv. ROE is used. Both dependent variables are constructed by rolling previous 12 quarters in order to persist with Z-score (12 quarters) which is also computed over. 政治大 Ratio to examine bank risk. 立In Model 7, LLA Ratio is taken place to test bank risk.. previous 12 quarters in baseline model (Model 1). Next, in Model 6 we employ NPL. ‧ 國. 學. Table 8 reports all results.. We find that all of the models are consistent with our previous finding suggests. ‧. that banks can lower risk with higher degree of internationalization; furthermore,. sit. al. er. io. at 1% level.. y. Nat. instead of Model 5 and Model 7, each of the specifications are statistically significant. n. Table 8 Alternative Measures of Risk This table reports regression analysis of the relationship between the internationalization of Taiwanese commercial banks and risk using alternative measures of risk to compare with. Ch. engchi. i n U. v. baseline model. Robust t-statistics are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels respectively. Dependent Variable: Alternative Measures of Risk (1) Z-score. (2) Z-score. (3) Z-score. (12 quarters). (8 quarters). (20 quarters). FAR. 1.526*** (7.74). 1.047*** (5.45). 2.085*** (9.61). Intercept. 3.100***. 3.676***. 2.826***. DIV. (10.62) -0.00028 (-0.60). (13.13) -0.00001 (-0.02). (8.75) -0.00040 (-0.95). Independent Variable. 19.

(26) Size. 0.032 (1.42). -0.003 (-0.15). 0.051** (2.06). Listed. -0.433*** (-8.65) 0.137** (2.47). -0.331*** (-6.77) 0.099 (1.81). -0.676*** (-12.94) 0.160* (2.81). OHC. -76.690*** (-16.54). -75.500*** (-17.47). -80.627* (-14.66). Time Fix Effect. Yes. Yes. Yes. Observations R-Square. 1,263 0.284. 1,502 0.233. 859 0.400. FHC. 0.228 治政 Measures Dependent Variable: Alternative of Risk (Conti.) 大 (5) (6) 立 (4) 0.278. -0.135*. (4.57). (-1.86). -1.392*** (-3.50) 0.00103. 0.659 (6.14) 0.00002. (1.61) 0.158*** (5.14) -0.009 (-0.14). (0.13) -0.046*** (-5.49) -0.025 (-1.35). -0.046*** (-0.61) -10.841*** (-1.72). Time Fix Effect Observations R-Square Adj. R-Square. Nat. DIV. io. Size. n. al. Listed FHC OHC. -0.077***. -0.006. (-6.53). (-1.27). 0.163*** 0.070*** (10.45) (11.26) -0.00015*** -0.00005*** (-4.84) -0.010*** (-7.88) -0.014*** (-4.70). (-4.29) -0.004*** (-8.59) -0.004*** (-3.45). 0.021 (1.02) 24.276*** (14.23). 0.0002 (0.06) 2.520*** (12.58). 0.003** (2.51) 0.732*** (9.21). Yes. Yes. Yes. Yes. 1,263 0.316 0.311. 1,263 0.229 0.223. 1,991 0.235 0.232. 1,991 0.142 0.138. Ch. engchi. 20. er. Intercept. (7) LLAR. y. 1.226***. FAR. NPLR. 0.393. ‧. Stdv. ROE. 學. RAROE. ‧ 國. Independent Variable. sit. Adj. R-Square. i n U. v.

(27) 4.4.2 Alternative measures of internationalization As noted, our main internationalization variable is Foreign Assets Ratio, but we replace it with Foreign Loans Ratio (Model 2) and Foreign Deposits Ratio (Model 3) to strengthen our research. As shown in Table 9, although the coefficient on Foreign Deposits Ratio does not reach statistically significant level, all coefficients on internationalization Ratios have the same positive impact on Z-score. This fact convinces that internationalization and risk have a negative relationship among banks in Taiwan.. 治政 This table reports regression analysis of the relationship between 大 the internationalization of Taiwanese commercial banks 立 and risk using alternative measures of internationalization to. Table 9 Alternative Measures of Internationalization. ‧ 國. 學. compare with baseline model. Robust t-statistics are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels respectively. (2) FLR. ‧. (3) FDR. 1.526***. 1.814***. 0.230. (7.74). (8.89). 3.100*** (10.62) -0.00028 (-0.60). 2.848*** (9.76) -0.00023 (-0.50). 3.175*** (10.63) -0.00011 (-0.23). 0.032 (1.42) -0.433*** (-8.65). 0.050** (2.23) -0.401*** (-8.05). 0.038 (1.63) -0.423*** (-8.26). 0.137** (2.47) -76.690*** (-16.54). 0.086 (1.54) -73.582*** (-15.83). 0.160*** (2.77) -81.233*** (-17.25). Time Fix Effect. Yes. Yes. Yes. Observations. 1,263. 1,263. 1,263. Nat. Internationalization Ratio. Size Listed FHC OHC. al. n. DIV. io. Intercept. Ch. er. Independent Variable. sit. (1) FAR. y. Dependent Variable: Z-score (12 quarters). n U engchi. 21. iv. (0.79).

(28) R-Square Adj. R-Square. 0.284 0.278. 0.294 0.289. 0.250 0.244. 4.5 Additional Analysis 4.5.1 Internationalization and risk during financial crisis Some argue that international banks are benefited from a more efficient asset diversification on global markets but, at the same time, exposing to more risk as well. To evaluate how does financial crisis impact on commercial banks in Taiwan, we create the. 政治大. subsample of subprime lending crisis from 2007:Q3 to 2009:Q4 and the results are presented. 立. in Table 10.. ‧ 國. 學. The results suggest that the banks in Taiwan during the financial crisis have a higher impact on Z-score compared to baseline model. According to the quarterly report of Bank of. ‧. Taiwan, Taiwanese banks did not hold many assets associated to subprime mortgage in 2007.. sit. y. Nat. Only 5 commercial banks sold related financial products to public investors with total amount. io. al. er. NTD 4.1 billion. Therefore, during financial crisis, banks still can reduce their risk through. n. portfolio diversification globally.. Ch. engchi. i n U. v. Table 10 Internationalization and Bank Risk during Financial Crisis This table reports regression analysis of the relationship between the internationalization of Taiwanese commercial banks and risk during subprime financial crisis. Robust t-statistics are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels respectively. Dependent Variable: Z-score (12 quarters) (1). (2). Independent Variable. Full Sample. Financial Crises. FAR. 1.526*** (7.74). 1.854*** (3.70). Intercept. 3.100***. 2.463***. DIV. (10.62) -0.00028. (3.37) -0.00054. 22.

(29) (-0.60) 0.032. (-0.97) 0.081. (1.42) -0.433*** (-8.65) 0.137**. (1.44) -0.831*** (-6.99) 0.153. OHC. (2.47) -76.690*** (-16.54). (1.18) -93.391*** (-8.67). Time Fix Effect. Yes. Yes. Observations. 1,263. 281. Size Listed FHC. R-Square Adj. R-Square. 政治大 0.284 0.278. 立. ‧ 國. 學. 4.5.2 Bank Size Break down. 0.386 0.366. In our previous studies, all evidences demonstrate that a negative correlation exists. ‧. between internationalization and risk; however, in our previous empirical analysis, the control. sit. y. Nat. variable, Size, has an ambiguous impact on risk without statistically significant. Accordingly,. io. er. we plot out the distribution of Size in Figure 3. As you can see from the figure, the distribution. al. of Size is bimodal which may have caused biases in regression models. To solve this problem,. n. v i n C has the cut point to U we select 70 percentile of Size (13.64) e n g c h i separate our sample into Small Size and Large Size which is the gray area in Figure 3. Next, we do the regression analysis to examine the relationship between internationalization and risk in two subsamples. In Table 11, we find that the banks in the sample of large size have a positive impact of internationalization on risk with statistically significant at 1% level. This result is consistent with Berger, Allen N., Sadok El Ghoul, Omrane Guedhami, and Raluca A. Roman (2013), who claim that the market risk hypothesis may dominate over the diversification hypothesis (i.e. banks will take on more risk when going abroad and cannot get the advantage on portfolio diversification). After separating the full sample into larger size and small size, we were 23.

(30) convinced that some of the banks characteristic variables are missing in our regression model. 250. Frequency. 200. 150. 100. 50. 立. 0 10.32. 10.90. 11.49. 12.07. 12.66. 13.24. 13.83. 14.41. 15.00. 學. Logarithm of Total Asset. ‧ 國. 9.73. 政治大. Figure 3 Distribution of Logarothm of Total Asset. ‧. Table 11 Regression Analysis by Different Bank Size. sit. y. Nat. This table reports regression analysis of the relationship between the internationalization of Taiwanese commercial banks and risk using difference bank size subsamples. Robust t-. n. al. er. io. statistics are in parentheses. ***, **, and * indicate significance at the 1%, 5%, and 10% levels respectively.. Ch. i n U. v. Dependent Variable: Z-score (12 quarters) (1). engchi. (2). (3). Independent Variable. Full Sample. Large Size. Small Size. FAR. 1.526*** (7.74). -1.085*** (-3.16). 1.871*** (7.56). Intercept. 3.100*** (10.62). -1.423 (-1.24). 4.884*** (8.94). DIV. -0.00028 (-0.60) 0.032 (1.42). 0.95138*** (3.61) 0.344*** (4.40). -0.00035 (-0.66) -0.122*** (-2.76). -0.433*** (-8.65). -0.305*** (-4.09). -0.264*** (-3.59). Size Listed. 24.

(31) FHC. 0.137** (2.47). 0.150*** (2.72). -0.020 (-0.21). OHC. -76.690*** (-16.54). -68.365*** (-6.24). -78.722*** (-14.21). Time Fix Effect. Yes. Yes. Yes. Observations. 1,263. 443. 820. R-Square Adj. R-Square. 0.284 0.278. 0.370 0.356. 0.291 0.283. In order to investigate bank characteristics in different bank size, we decompose Z-score into three main components which are mean ROA, mean capitalization asset ratio (Equity/Total. 政治大. asset), and standard deviation of return on asset (Stdv. ROA). In Table 12, we find that large. 立. banks have higher average Z-score, lower capitalization asset ratio, and lower Stdv. ROA than. ‧ 國. 學. small banks which mean although large banks tend to take on more risk from internationalization activates, large banks are more stable than small banks. Furthermore, large. ‧. banks are associated with lower profitability than small banks as measured by mean ROA.. y. Nat. sit. Table 12 Z-score Decomposition by Different Bank Size. n. al. er. io. This table reports Z-score decomposition for large size versus small size. ***, **, and * indicate significance at the 1%, 5%, and 10% levels respectively.. Ch. i n U. Z-score Decomposition. Large Size N=443. e n g Small c h i Size N=820. v. Large - Small (Mean). Variable. (1) Mean. (2) Mean. (3) Diff.. (4) t-stat. Z-score (12 quarters) ROA. 3.4269 0.0039. 3.1769 0.0070. 0.2500 -0.0031. 5.89*** -9.12***. Equity/Asset Stdv. ROA. 0.0575 0.0024. 0.0977 0.0064. -0.0402 -0.0040. -10.79*** -8.70***. 25.

(32) 5 Conclusions Many research papers show that the more internationalized a bank is, the higher the risk it bears. However, based on commercial banks data in Taiwan, we find that internationalization has negative relationship with risk following Berger, A., Sadok El Ghoul, Omrane Guedhami, and Raluca A. Roman (2013) structures, who analyze the relationship between internationalization and bank risk with numerous of empirical models. Also, we deliver the robustness test on different measures of risk and internationalization suggesting a consistent result to support our finding. In the end, we consider banks in distinct size because we find a. 政治大 with 70p cut point, a strong evidence 立 shows that large banks tend to take on more risk but small bimodal distribution in Size variable. After we subsample the data into large size and small size. ‧ 國. 學. banks can be benefited by diversify their portfolios and gain access to global capital market. We show large banks could gain more risk from increasing internationalization activities.. ‧. However, when we compare characteristics of bank between the two subsamples we find that. sit. y. Nat. large bank have a higher average Z-score and higher income diversification indicating that. n. al. er. io. large banks can gain more profit and be more stable. On the other hand, small banks have a. i n U. v. diversification effect on internationalization but the overall risk is larger and the operation is. Ch. engchi. not as efficient as large banks. The reason is that regulations in Taiwan are more restricted for banks inside of the country which make only several financial holdings can do financial services differently in Asia Pacific; therefore for those who seeking more profits outside Taiwan might take on more risk. Conversely, the small banks can also provide financial services which are relatively simple and no risky; as a consequence, they could benefit from asset diversification. We suggest that different bank sizes may have different impacts of internationalization on risk and other evaluation methods should be employed in different banks.. 26.

(33) Reference 1. Aliber, Robert Z., 1976. ‘International bank: A survey,’ Journal of Money, Credit, and Banking, 16(25), 47-51. 2. Amihud, Y., DeLong, G. and Saunders, A., 2002. ‘The effects of cross-border bank mergers on bank risk and value,’ Journal of International Money and Finance, 21, 857-877. 3. Ana Rosa Fonseca and Francisco González, 2010. ‘How bank capital buffers vary across countries: The influence of cost of deposits, market power and bank regulation,’ Journal of Banking & Finance, 34(4), 892-902.. 政治大. 4. Annavarjula, M. and Beldona, S., 2000. ‘Multinationality performance relationship: A. 立. review and reconceptualization,’ International Journal of Organizational Analysis, 8(1),. ‧ 國. 學. 48-67.. 5. Berger, A. and Bouwman, C., 2013. ‘How does capital affect bank performance during. ‧. financial crises?’ Journal of Financial Economics, 109(1), 146-176.. y. Nat. sit. 6. Berger, A., Bouwman, C., Kick, T. and Schaeck, K., 2012. ‘Bank risk taking and liquidity. n. al. er. io. creation following regulatory interventions and capital support,’ Working paper, University of South Carolina.. Ch. engchi. i n U. v. 7. Berger, A., Sadok El Ghoul, Omrane Guedhami and Raluca A. Roman, 2013. ‘Internationalization and bank risk,’ Working paper, University of South Carolina. 8. Bhagat S., Bolton, B. and Lu J., 2012. ‘Size, leverage, and risk-taking of financial institutions,’ Working paper, University of Colorado at Boulder. 9. Boyd, J. and Runkle, D., 1993. ‘Size and performance of banking firms,’ Journal of Monetary Economics, 31(1), 47-67. 10. Buch, Claudia M., John C. Driscoll and Charlotte Ostergaard, 2010. ‘Cross-border diversification in bank asset portfolios,’ International Finance, 13(1), 79-108.. 27.

(34) 11. Cetorelli, N. and Goldberg, L., 2012. ‘Banking globalization and monetary transmission,’ Journal of Finance, 67, 1811-1843. 12. Chiorazzo, V., Milani, C. and Salvini, F., 2008. ‘Income diversification and bank performance: Evidence from Italian banks,’ Journal of Financial Services Research, 33, 181-203. 13. Demirgüç-Kunt, A. and Huizinga, H., 2010. ‘Bank activity and funding strategies: The impact on risk and returns,’ Journal of Financial Economics, 98, 626-650. 14. Demsetz, R. S. and Strahan, P. E., 1997. ‘Diversification, size and risk at bank holding. 政治大. companies,’ Journal of Money, Credit, and Banking, 29 (3), 300-313.. 立. 15. Doukas, J. and Kan, O. 2006. ‘Does global diversification destroy firm value?’ Journal of. ‧ 國. 學. International Business Studies, 37, 352-71.. 16. Flamini, V., McDonald, C.A. and Schumacher, L.B., 2009. ‘The determinants of. ‧. commercial bank profitability in Sub-Saharan Africa,’ IMF Working Paper, No. 09/15.. y. Nat. io. sit. 17. Giuliano Iannottaa, Giacomo Nocerab, and Andrea Sironi 2013. ‘The impact of. n. al. er. government ownership on bank risk,’ Journal of Financial Intermediation, 22(2), 152-176.. i n U. v. 18. Grubel, Herbert G., 1977. ‘A theory of multinational banking,’ Banca Nazionale del Lavoro Quarterly Review, 349-363.. Ch. engchi. 19. Gulamhussena, M.A., Pinheirob, C. and Pozzolo A.F., 2014. ‘International diversification and risk of multinational banks: Evidence from the pre-crisis period,’ Journal of Financial Stability, 13, 30-40. 20. Houston, Joel F., Chen Lin, Ping Lin and Yue Ma., 2010. ‘Creditor rights, information sharing, and bank risk-taking,’ Journal of Financial Economics, 96(3), 485-512. 21. Stiroh, K. J. and Rumble, A., 2006. ‘The dark side of diversification: The case of US financial holding companies,’ Journal of Banking & Finance, 30(8), 2131-2161.. 28.

(35) 22. Laeven, L. and Levine, R., 2007. ‘Is there a diversification discount in financial conglomerates?’ Journal of Financial Economics, 85, 331-67. 23. Laeven, L. and Levine, R., 2009. ‘Bank governance, regulation and risk taking,’ Journal of Financial Economics, 93, 259-275. 24. Laeven, L., Ratnovski, L. and Tong, H., 2014. ‘Bank size and systemic risk,’ International Monetary Fund. 25. Mamiza Haq and Richard Heaney, 2012. ‘Factors determining European bank risk,’ Journal of International Financial Markets, Institutions and Money, 22(4), 696-718.. 政治大. 26. Shen, C. H. and Chang Y., 2012. ‘To join or not to join?-Do banks that are of a Financial-. 立. Holding-Company perform better than banks that are not,’ Contemporary Economic Policy,. ‧ 國. 學. 30(1), 113-128.. 27. Thi Canh Nguyen, Dinh Vinh Vo and Van Chien Nguyen, 2015. ‘Risk and income. ‧. diversification in the Vietnamese banking system,’ Journal of Applied Finance & Banking,. sit. y. Nat. 5(1), 93-109.. n. al. er. io. 28. Wang, Tianxi, 2014. ‘Competition and Increasing Returns to Scale: A Model of Bank Size’ The Economic Journal, Early View.. Ch. engchi. 29. i n U. v.

(36)