The Relationship and Causality Testing between Diversification, Risk ...

The Relationship and Causality Testing between Diversification, Risk, and Financial Performance: Empirical Examination in Taiwan’s Banking Industry

Shu Ling Lin1 Soushan Wu2 Jack H.W. Penm3

Submitted to Asian FA/TFA/FMA 2003 Conference

28 February 2003

1

Associate professor, Department of International Trade and Finance, Fu-Jen Catholic University. Address: 510 Chungchen Rd., Hsinchuang, Taipei Hsien, Taiwan 242, R.O.C., Phone: 886-2-29031111 ext. 2721, Fax: 886-2-22027979, Email: [email protected].

2

Corresponding author. Dean and Professor, College of Management, Chang Gung University, Tao Yuan, Taiwan, Email: [email protected].

3

Professor, Faculty of Economics and Commerce, The Australian National University, Canberra, Australia. Address: 9 NEEDHAM PLACE STIRLING A.C.T. 2611 AUSTRALIA. Email: [email protected]

Abstract The objective of this study is to examine the portfolio theory that suggests that diversification can potentially reduce the return variance and the probability of failure of a portfolio. The current study is to investigate the diversification measure by means of Hershman-Herfindahl index, and applies the “diversification index” proposed by Demsetz & Strahan (1997) by scaling systematic risk with stock return variance of individual bank, which implies that as the banking with a high diversification index (a high R2), the fraction of risk stemming from firm-specific factors is small. Next, this study examines the effects of diversification on risk and financial performance respectively, and finally tests the causality between diversification, risk, and financial performance in Taiwan’s banking industry during 1993~2001. The empirical results shows that the average diversification measures based on Hershman-Herfindahl index between stated-owned banking and new-private banking are 26.61% and 25.18% respectively. It shows that the state-owned banking is better diversified than new-private banking. However, better diversification does not translate into reductions in risk for state-owned banking. Besides, this study also demonstrates a positive relationship between systematic risk and stock return variance of banking industry in Taiwan. Furthermore, the empirical results show that the more diversification, the less volatility of ROA and possibility of failure. In addition, the effects of diversification on five kinds of financial performance show that the more of diversification, the more of ROA, EPS, NIS, PIS, and ROE. The results show that the diversification and the financial performance of banking industry exist positive relationship significantly; it implies that the diversity nonbank activities are benefited to the financial performance of banking industry. On the side of the relationship between risk and financial performance, the empirical results show that the return variance and the five kinds of financial performances exists negative relationship significantly. That is, the results show that the more of market risk, the less of financial performance for banking industry. The results of Granger causality testing between diversification and risk shows that the diversification of past period does cause the risks of the banking industry; there

i

exist the time lag effect between diversification and risk. Besides, the results of Granger causality testing between diversification and financial performance shows that causality testing between diversification and five kinds of financial performance shows that the diversification of past period does cause the financial performance of the banking industry, and vice versa; there exist the time lag effect between diversification and financial performance. This study provides strong evidence of a link between diversification, risk, and financial performance of banking industry in Taiwan. The empirical examinations suggest that diversification may provide an important motive for risk reducing and performance enhanced for banking industry.

Keywords: Diversification, the Possibility of Failure, Capital Adequate Ratio (BIS), Return on Assets (ROA), Return on Equity (ROE), Earnings Per Share (EPS).

ii

I. Introduction The global financial environment is experiencing significant change over the past two decade. More recently, a number of countries have experienced banking crises and undertaken significant regulatory change and geographical barriers to competition are reduced. The rapidly changing environment for financial institutions has resulted in increasing research concerned with exploring the behavior and performance of banking industry. In recent years, much research attention has been focused to examination and comparative analyses of banking regulation, structure, behavior, risk and performance. As the banking industry becomes more global and integrates, the nonbank activities induce diversification has risk-reducing potential. Thus, the importance of diversification translates into risk reductions resulting in different performance accelerate. Applied the fundamental implication of modern portfolio theory to banking, portfolio theory suggests that diversification can potentially reduce the return variance and the probability of failure of a portfolio. The banking literature tends to assume that diversification, risk, and financial performance go hand in hand. This study demonstrates empirically that this presumption is valid. The objectives of this study including the follows: Firstly, this study investigates the measure of diversification by means of Hershman-Herfindahl index, that is, the ratio of variety activities over the bank’s total revenues. Secondly, this study applies the “diversification index” proposed by Demsetz & Strahan (1997) by scaling systematic risk with stock return variance, which implies that while the banking with a high diversification index (a high R2), the fraction of risk stemming from firm-specific factors is small. Thirdly, this study examines the effects of diversification on risk and financial performance, and finally tests the causality between diversification, risk, and financial performance in Taiwan’s banking

industry

during

1993~2001.

The

ordinary

least

square

(OLS)

multiple-regression is applies for examination the effects of diversification on risk and financial performance. Besides, a robust estimation by Yule-Walker estimates for error term autocorrelations is used to control the potential problems caused by outliers. Furthermore, the weighted least square (WLS) regression for heteroscedasticity variance of the error term is also performed. Finally, the Granger causality testing is used to test

1

the time-lag effect between diversification, risk, and financial performance. The structure of this study arranged as follows: Section II develops the hypotheses to be tested for this study. Section III describes the empirical model and statistic measure. Section IV presents the empirical results. Finally, the empirical results’ implicat ions are concluded.

II. Hypotheses Development and Empirical Model From the above-mentioned literature and inferences, the objectives of this study is to investigate the diversification measure by means of Hershman-Herfindahl index, the ratio of variety activities over the bank’s total revenues. Subsequently, this study applies the “diversification index” proposed by Demsetz & Strahan (1997) by scaling systematic risk with stock return variance of individual bank, while the banking with a high diversification index (a high R2), the fraction of risk stemming from firm-specific factors is small. After that, this study examines the effects of diversification on risk and financial performance

respectively, and finally

tests

the

causality between

diversification, risk, and financial performance in Taiwan’s banking industry during 1993~2001. The above objectives would be obtained by testing the following alternative hypotheses: To investigate the relationships and causality testing between diversification, risk, and financial performance of Taiwan’s Stated -Owned banking and New-Private banking, the empirical models to be estimated in this study to test hypothesis 1~4 are proposed as follows.

2.1 The Relationship between Diversification, Risk, and Financial Performance Hypothesis 1: The relationship between market risk and total risk of individual bank is positive. Specifically, while the banking with a high diversification index (a high R2), the fraction of risk stemming from firm-specific factors is little. Thus, more of market risk

2

implicitly represents less of firm-specific risk. The empirical model to be estimated to test hypothesis 1 is proposed as follows:

σ 2 (Rb ) = β 2 × σ 2 (Rm ) + σ 2 (ε ) Where σ

2

(Rb )

(1)

means the variance of individual bank stock return, which is represented the total risk of individual bank.

σ 2 (Rm ) means the variance of Taiwan’s weighted stock return, which is represented the undiversifiable risk.

σ 2 (ε ) means the firm-specific risk, which is represented the diversifiable risk. Hypothesis 2: The relationship between diversification measure and risk is negative. That is, the more of diversification in banking will reduce the operating risk, which implicitly represents higher diversification measure would induce lower operating risk by the banking. The empirical models to be estimated to test hypothesis 2 are proposed as follows:

β = α 0 + α1 (DIV ) + α 2 (SIZE ) + α 3 (Z ) + ε

(2)

SD (ROA) = β 0 + β1 (DIV ) + β 2 (SIZE ) + β 3 (Z ) + ε

(3)

= γ 0 + γ 1 (DIV ) + γ 2 (SIZE ) + γ 3 (Z ) + ε

(4)

BIS

Where β means the systematic risk. DIV means the measure of diversification. Size means the asset size of the bank. Z means the dummy variable for the Stated-Owned bank and the New-Private bank respectively, while Z=0 refer to Stated-Owned bank, while Z=1 means New-Private bank.

3

ε means the error term. SD (ROA) means the standard deviation of return on assets. BIS means the capital adequately ratio. Hypothesis 3: The relationship between diversification measure and financial performance is positive. Specifically, the more of diversification in banking will induce the higher of financial performance (ROA, ROE, NIS, PIS, and EPS), which implicitly represents higher diversification measure would induce lower financial risk, and then improve the financial performance by the banking. The empirical models to be estimated to test hypothesis 3 are proposed as follows:

ROA = δ 0 + δ 1 (DIV ) + δ 2 (SIZE ) + δ 3 (Z ) + ω 0

(5)

ROE = ζ 0 + ζ 1 (DIV ) + ζ 2 (SIZE ) + ζ 3 (Z ) + ω 0

(6)

NIS = η0 + η1 (DIV ) + η 2 (SIZE ) + η3 (Z ) + ω 0

(7)

PIS = θ 0 + θ1 (DIV ) + θ 2 (SIZE ) + θ 3 (Z ) + ω 0

(8)

EPS = ι0 + ι1 (DIV ) + ι 2 (SIZE ) + ι3 (Z ) + ω 0

(9)

Where ROA means the return on assets of the bank. ROE means the return on equities of the bank. NIS means the ratio of net income over its total revenues of the bank. PIS means the ratio of EBIT (earnings before interest and tax) over its total revenues of the bank. EPS means the ratio of net income over its total outstanding shares of the bank.

ω means the error term. Hypothesis 4: The relationship between risk and financial performance is negative.

4

That is, the more of diversification in banking will induce the less of operating risk and the higher of financial performance (ROA, ROE, NIS, PIS, and EPS) that implicitly represents higher diversification measure would induce lower financial risk, and then improve the financial performance by the banking. The empirical models to be estimated to test hypothesis 4 are proposed as follows:

ROA = κ 0 + κ 1 (SD (ROA)) + κ 2 (SIZE ) + κ 3 (Z ) + ε

(10)

ROE = λ0 + λ1 (SD (ROA)) + λ2 (SIZE ) + λ3 (Z ) + ε

(11)

NIS = µ 0 + µ1 (SD (ROA)) + µ 2 (SIZE ) + µ 3 (Z ) + ε

(12)

PIS = ν 0 + ν 1 (SD (ROA)) + ν 2 (SIZE ) + ν 3 (Z ) + ε

(13)

EPS = ξ 0 + ξ1 (SD (ROA)) + ξ 2 (SIZE ) + ξ 3 (Z ) + ε

(14)

ROA = ο 0 + ο1 (BIS ) + ο 2 (SIZE ) + ο 3 (Z ) + ε

(15)

ROE = π 0 + π 1 (BIS ) + π 2 (SIZE ) + π 3 (Z ) + ε

(16)

NIS = ρ 0 + ρ1 (BIS ) + ρ 2 (SIZE ) + ρ 3 (Z ) + ε

(17)

PIS = σ 0 + σ 1 (BIS ) + σ 2 (SIZE ) + σ 3 (Z ) + ε

(18)

EPS = τ 0 + τ 1 (BIS ) + τ 2 (SIZE ) + τ 3 (Z ) + ε

(19)

ROA = ν 0 + ν 1 (β ) + ν 2 (SIZE ) + ν 3 (Z ) + ε

(20)

ROE = ϕ 0 + ϕ 1 (β ) + ϕ 2 (SIZE ) + ϕ 3 (Z ) + ε

(21)

NIS = χ 0 + χ1 (β ) + χ 2 (SIZE ) + χ 3 (Z ) + ε

(22)

PIS = ψ 0 + ψ 1 (β ) + ψ 2 (SIZE ) + ψ 3 (Z ) + ε

(23)

5

EPS = ω 0 + ω1 (β ) + ω 2 (SIZE ) + ω 3 (Z ) + ε

(24)

Acceptation of the above alternative hypotheses would support the argument that there exists a relationship between diversification, risk, and financial performance in Taiwan’s banking industry. The sign of coefficient will indicate whether the relationship is positive or negative.

2.2 The Causality Testing between Diversification, Risk, and Financial Performance In order to test for causality between diversification, risk, and financial performance, this study selects the Granger’s (1969) causality testing. This study considers the following simple structural model for the causality testing between diversification and risk; the forms of bivariate regressions are as follows:

Riskt = α 0 + α1Riskt −1 + ...... + α l Riskt − l + β1Divt −1 + ..... + β l Divt − l + ε t Divt = α 0 + α1Divt −1 + ...... + α l Divt − l + β1 Riskt −1 + ..... + β l Riskt − l + µ t .(25) Where Risk means the risk of Beta, sd (ROA), and BIS respectively. Div means the measure of diversification. For the above all possible pairs of (Div, Risk) series in the group, the reported F-statistics are the Wald statistics for the joint hypothesis as follows:

β1 = β 2 = β 3 = .......... = β l = 0 For the above equation, the null hypothesis is that diversification does not Granger-cause Risk in the first regression and that Risk does not Granger-cause diversification in the second regression. Besides, this study considers the following simple structural model for the causality testing between diversification and financial performance; the forms of bivariate regressions are as follows:

FPt = α 0 + α1 FPt −1 + ...... + α l FPt − l + β1Divt −1 + ..... + β l Divt − l + ε t

6

Divt = α 0 + α1Divt −1 + ...... + α l Divt − l + β1FPt −1 + ..... + β l FPt − l + µ t

(26)

Where FP means the financial performance of ROA, ROE, NIS, PIS, and EPS respectively. In addition, for the above all possible pairs of (Div, FP) series in the group, the reported F-statistics are the Wald statistics for the joint hypothesis as follows:

β1 = β 2 = β 3 = .......... = β l = 0 For the above equation, the null hypothesis is that diversification does not Granger-cause Financial Performance in the first regression and that Financial Performance does not Granger-cause diversification in the second regression. The next section discusses the empirical model and sample selection.

III. Statistic Measure 3.1 Sampling Selection and Study Period The sample of corporations for this study is drawn from the publicly traded securities in Taiwan’s banking industry over the period of 1993 and 2001. Only those bank which is publicly in 1993 are selected since the main research variable are most salient to these corporations. After applying the screen to the population, 35 banks are identified from Taiwan’s banking industr y in the study sample; including 32 publicly banks on the Taiwan Stock Exchange, and 3 banks on the OTC market. The study sample is further divided into two group by the deregulation of banking establish form in 1991. Consequently, the one of the sample group is stated-owned banking, the others is new-private banking. (Appendix 1) Besides, all data for the analysis is extracted from the sample year-end financial statements from Taiwan Economic Journal (TEJ) over the period 1993 to 2001.

3.2 Variable definition The dependent variables include risk and financial performances in this study. The 7

risk measures comprise the Beta, sd (ROA), and BIS. Additionally, the financial ratios including Return on Assets (ROA), Return on Equity (ROE), Net Income over Total Revenues (NIS), EBIT over Total Revenues (PIS), and Earnings per Share (EPS) for each bank which are collected and derived from the publicly statement for those sample corporations for the years 1993-2001. The independent variable, diversification measures (DIV) that are calculated by the ratio of variety activities over the banks total revenues by means of Hershman-Herfindahl index. Subsequently, this study applies the “diversification index” proposed by Demsetz & Strahan (1997) by scaling systematic risk with stock return variance of individual bank, while the banking with a high diversification index (a high R2), the fraction of risk stemming from firm-specific factors is small. The formulas used for defining diversification, risk, and financial performance variables, are summarized as follows:

3.2.1 Independent Variable: Diversification This study proposes the diversification measure by means of the ratio of diversify operating revenue over its total revenues. If the diversification is high, then the ratio of diversification measure is far above the ground too. This is implies that the diversification of operating activity of the bank is high. The formula used for defining diversification variable that is summarized as follows: DIV

n

= 1 − ∑ Pi 2

(27)

i=1

Where DIV means the measure of diversification.

Pi means the ith operating revenue over total revenue of the bank. n means the item number of operating activity of the bank, which including interest income, fees income, gain from sales of note, gain from sale of long-term equity investment, gain from currency exchange, and other income.

3.2.2 Dependent Variable 8

3.2.2.1. Risk Measure: (1) Market Risk (Beta ) : By scaling systematic risk of Taiwan’s Weighted Stock Index with stock return variance of indivivual bank,

(2) Standard Deviation of ROA ( sd ( ROA)) : This study uses the standard deviation of ROA to measure the soundness of the sample banks. 9

∑ ( ROAi , t − ROAi , t ) 2

sd ( ROAi , t ) = t =1

n −1

(28)

(3) Capital Adequately Ratio (BIS): In order to measure the stableness and soundness, this study adopts the BIS ratio of the sample banks.

BIS =

NetCapital RiskyAsset

(29)

3.2.2.2 The Measures of Financial Performance: (1) Return on Assets (ROA): ROA =

NetIncome TotalAssets

(30)

NetIncome TotalEquity

(31)

(2) Return on Equity (ROE): ROE =

(3) Ratio of Operating Income (NIS):

9

NIS =

( Interest + Service) Income Total Re venue

(32)

(4) Ratio of EBIT (PIS): PIS =

EBIT Total Re venue

(32)

(5) Earnings per Share (EPS): EPS =

NetINcome Outs tan dingShares

(33)

3.2.3 Control Variables: According to Wernerfelt (1984) and Ghemawat (1991), since the performance variables are not standardized, it is necessary to include the control variables to blunt the effect of differences in corporation commitments, resources, and strategies. As a result, in order to control for these potentially value-relevant cross-sectional differences, the analysis includes the following control variables: bank size (log (Total assets)), and dummy variable for the stated-owned bank and the new-private one respectively.

(1) Bank Size (Size): In order to determine the size of the sample banks, this study adopts the log of total assets measure the size.

Size = log(Assets)

(34)

(2) Dummy Variable for bank (Z): Z=0, as the bank belongs to stated-owned banking. Z=1, as the bank belongs to new-private banking.

3.3 Statistic method The ordinary least square (OLS) multiple-regression is applies for examination the effects of diversification on risk and financial performance. Besides, a robust

10

estimation by Yule-Walker estimates for error term autocorrelations is used to control the potential problems caused by outliers. Furthermore, the weighted least square (WLS) regression for heteroscedasticity variance of the error term is also performed. Finally, the Granger causality testing is used to test the time-lag effect between diversification, risk, and financial performance during 1993 to 2001.

IV. Empirical Results 4.1 Descriptive statistics The descriptive statistics of the entire group of samples indicate that the standard deviation of ROE, NIS, and PIS are large, which are 16.7494%, 23.4785%, and 14.6191% respectively. This shows that the sample corporations’ change of return to shareholders between 1993-2001 period is tremendous. Besides, the mean value of the variance of market return is negative, whereas the standard deviation is 0.0435. Because the value of Bank Size measurement error is between 8.7307 and 12.1059, the measurement error of Size among the sample banks is quite significant. This indicates that the control variable in the model is appropriate. Insert Table 1 Here

4.2 The Measure of Diversification The empirical results shows that the average diversification measures based on Hershman-Herfindahl index for stated-owned banking is 26.61%, and 25.18% for new-private banking during 1993~2001. It shows that the stated-owned banking is better diversified than new-private banking. However, better diversification does not translate into reductions in risk for stated-owned banking, the standard deviation is 10.13% and 8.39% respectively. Insert Table 2 Here This study also demonstrates a positive relationship between systematic risk and stock return variance of banking industry; the adjusted R2 shows 20.47% which is implied that the 79.53% of firm-specific risk is diversified away by diversification. 11

Insert Table 3 Here

4.3 Multiple Regression Analysis on the Relationship between Diversification, Risk, and Financial Performance 4.3.1 The Effects of Diversification on Risk On the effects of diversification on risk, measured by beta, standard deviation of ROA, and capital adequate (BIS) ratio. The empirical results show that the more diversification, the less volatility of ROA and possibility of failure. The estimate coefficients are –1.6024 (p-value is 0.0001) and 13.4450 (p-value is 0.0001) respectively. The results show that there exists significantly negative relationship between diversification and risk. That is, diversification would reduce the risk of banking in Taiwan. Insert Table 4 Here

4.3.2 The Effects of Diversification on Financial Performance Besides, the effects of diversification on five kinds of financial performance show that the more of diversification, the more of ROA, EPS, NIS, PIS, and ROE. The estimate coefficient of diversification on ROA is 6.5779 (p-value is 0.0001), the EPS is 2.3981 (p-value is 0.0074), the NIS is 44.4155 (p-value is 0.0001), the PIS is 55.9626 (p-value is 0.0001), and the ROE is 35.0218 (p-value is 0.0005) correspondingly. The results show that the diversification and the financial performance of banking industry all exist positive significantly relationship; it implies that the variety nonbank activities are benefited to the financial performance of banking industry. That is, diversification would enhance the financial performance of banking in Taiwan. Insert Table 5 Here

4.3.3 The Relationship between Risk and Financial Performance On the side of the relationship between risk and financial performance, the

12

empirical results show that the increasing of return variance, the five kinds of financial performances are all reducing. That is, the standard deviation of ROA and five kinds of financial performances all exist negative significantly relationship. The estimate coefficient of return variance on the ROA, ROE, EPS, PIS, and NIS are as follows: –0.9646 (p-value is 0.0001), –9.0080 (p-value is 0.0001), –0.2913 (p-value is 0.0413), –14.0232 (p-value is 0.0001), and –9.6385 (p-value is 0.0001) correspondingly. That is, the lower of return variance would enhance the financial performance of banking in Taiwan. Insert Table 6 Here Furthermore, the BIS ratio and the five kinds of financial performances all show positive significantly relationship. The estimate coefficient of BIS ratio on ROA, ROE, EPS, PIS, and NIS are as follows: 0.0868 (p-value is 0.0001), 0.6427 (p-value is 0.0002), 0.0596 (p-value is 0.0001), 1.5584 (p-value is 0.0001), and 0.6355 (p-value is 0.0001). The results imply that the more of capital adequate ratio, the banking will pursue less risky lending as operating with less leverage, then the less possibility of failure for banking. As a result, their financial performance will be enhanced. That is, the higher the BIS ratio would enhance the financial performance of banking in Taiwan. Insert Table 7 Here Finally, the empirical examinations show that the market risk (Beta) and ROE, NIS exist negative significantly relationship. The estimate coefficient of Beta on ROE, and NIS are –4.2214 (p-value is 0.0535) and –4.5353 (p-value is 0.251) respectively. The results show that the more of market risk, the less of financial performance for banking industry. That is, the higher the market risk would reduce the financial performance of banking in Taiwan. Insert Table 8 ~ 11 Here

4.4 The Causality Testing between Diversification, Risk, and Financial Performance 4.4.1 The Causality between Diversification and Risk 13

The results of Granger causality testing between diversification and risk are summarized as follows: The result of Granger causality testing between market risk (beta) and diversification shows that the past fourth year of market risk (beta) does cause diversification at the 10% significant level (the p-value is 0.096). Insert Table 12 Here The result of Granger causality testing between return variance (sd (ROA)) and diversification shows that the past first, second, third, and fourth year of diversification does cause the variance of ROA at the 1% significant level (the p-value are all show 0.0001). Insert Table 13 Here The result of Granger causality testing between BIS ratio and diversification shows that the past second, third, and fourth year of diversification does cause the BIS ratio at the 5% significant level (the p-value is 0.018, 0.041, and 0.05 respectively). Insert Table 14 Here Accordingly, the diversification of past period does cause the operating risk of the banking industry; That is, there exist the time lag effect between diversification and risk in Taiwan.

4.4.2 The Causality between Diversification and Financial Performance The results of Granger causality testing between diversification and financial performance are as follows: The result of Granger causality testing between diversification and ROA shows that the past first, second, third, and fourth year of diversification does cause the ROA at 5% significant level (the p-value is 0.061, 0.015, 0.026, and 0.044 respectively). Alternatively, the result shows that the past second, third, and fourth year of ROA does cause the diversification at 1% and 5% significant level (the 14

p-value is 0.009, 0.019, and 0.017 respectively). Insert Table 15 Here The result of Granger causality testing between diversification and ROE shows that the past first, second, third, and fourth year of diversification does cause the ROE at 1% significant level (the p-value are 0.003, 0.0001, 0.0001, and 0.001 respectively). Insert Table 16 Here The result of Granger causality testing between diversification and NIS shows that the past second, third, and fourth year of diversification does cause the NIS at 10% significant level (the p-value are 0.052, 0.081, and 0.079 respectively). Insert Table 17 Here The result of Granger causality testing between diversification and PIS shows that the past second, third, and fourth year of diversification does cause the PIS at 1% and 5% significant level (the p-value are 0.009, 0.019, and 0.017 respectively). Insert Table 18 Here The result of Granger causality testing between diversification and EPS shows that the past first, second, third, and fourth year of diversification does cause the EPS at 1% significant level (the p-value are 0.003, 0.005, 0.001, and 0.017 respectively). On the other hand, the result shows that the past second, third, and fourth year of EPS does cause the diversification at 1% significant level (the p-value are 0.009, 0.001, and 0.003 respectively). Insert Table 19 Here Consequently, the diversification of past period does cause the financial performance of the banking industry, and vice versa; That is, there exist the time lag effect between diversification and financial performance.

15

V. Concluding Remarks The empirical result of this study shows that the stated-owned banking is better diversified than new-private banking in Taiwan during 1993~2001. Besides, the result demonstrates a positive relationship between systematic risk and stock return variance of banking industry; it implies that diversification would diversify the firm-specific risk. This study provides strong evidence of a link between diversification, risk, and financial performance of banking industry in Taiwan. Specifically, there exists significantly negative relationship between diversification and risk. That is, diversification would reduce the operating risk of banking in Taiwan. Besides, Besides, The results show that the diversification and the financial performances of banking industry exist positive significantly relationship; it implies that diversification would enhance the financial performance of banking in Taiwan. Finally, the higher the operating risk would reduce the financial performance of banking in Taiwan. In addition, the results of Granger causality testing shows that diversification of past period does cause the operating risk and financial performance of the banking industry; That is, there exist the time lag effect between diversification, risk, and financial performance in Taiwan’s banking industry. The empirical examinations suggest that diversification may provide an important motive for risk reducing and performance enhanced for banking industry.

16

Table 1: Summary Statistics Variable

Number of Mean Observation

Standard Minimum Maximum Deviation

Symbol

Variance of stock return over the banking in our sample

174

0.0114

0.1231

-0.6156

0.6156

2

Variance of market return

174

-0.0080

0.0435

-0.0695

0.0851

2

Market risk

224

0.7753

0.6310

-3.9900

3.1542

BETA

Standard deviation of ROA Capital adequately ratio Diversification measure

312

0.5848

0.6482

0.0873

2.6106

SD (ROA)

309

13.2211

6.7704

-20.7700

43.5000

BIS

312

0.2591

0.0948

0.0365

0.5251

DIV

Assets Size

312

11.3065

0.4327

8.7307

12.1059

SIZE

Dummy variable for Stated-Owned and New-Private banking

312

0.4904

0.5007

0

1

Z

Return on assets

312

0.4636% 2.3220% -14.0300% 27.1800%

ROA

Return on equity

312

4.0349% 16.7494% -184.65% 38.5500%

ROE

Net Income / Sales

312

6.8339% 23.4785% -299.53% 74.0800%

NIS

EBIT / Sales

312

6.2791% 14.6191% -97.3000% 66.9200%

PIS

Net Income / Outstanding shares

312

$1.1926

EPS

$1.5912 $-10.7900

ó(Rb) ó(Rm)

$6.6200

Table 2: The Diversification Measures of Banking Industry in Taiwan Year

1993

1994

1995

1996

1997

1998

1999

2000

2001

Average

Stated-Owned

Mean

0.2541 0.2772 0.2653 0.2964 0.3097 0.2654 0.2851 0.2049 0.2370

0.2661

Banking

Standard deviation

0.1154 0.1099 0.1131 0.0967 0.0860 0.0998 0.0973 0.0915 0.1064

0.1013

Year

1993

1994

1995

1996

1997

1998

1999

2000

2001

Average

New-Private

Mean

0.3046 0.2311 0.2019 0.2558 0.2790 0.2397 0.2643 0.2211 0.2683

0.2518

Banking

Standard

0.0944 0.0666 0.0747 0.0474 0.0653 0.0521 0.0853 0.0998 0.1138

0.0839

deviation

17

Table 3: The Relationship between Market Return Variance and Stock Return Variance of Bank in Taiwan’s Banking Industry (1993~2001) Yuler-Walker Estimates

Stock Return Variance of Individual Bank in Taiwan Variance of Market Return

Regression parameter

P-value

1.4380

(0.0001)***

F-value (P-value)

62.266(0.0001)***

Adjusted R-square

0.2047

Durbin-Waston value

1.9648

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value.

Table 4: The Relationship between Diversification and Risk in Taiwan’s Banking Industry (1993~2001) Dependent Variable Statistical Method

Market Risk (beta)

Standard Deviation of ROA (sd (ROA))

Yuler-Walker Estimation

OLS

Capital Adequately Ratio (BIS) Yuler-Walker Estimation

Regression parameter

P-value

Regression parameter

P-value

Regression parameter

P-value

Diversification

-0.0372

(0.9250)

-1.6024

(0.0001)***

13.4450

(0.0001)***

SIZE

0.0001

(0.9995)

-0.5761

(0.0001)***

-7.4364

(0.0001)***

Dummy variable for Stated-Owned and New-Private banking

-0.5632

(0.0001)***

-0.1379

(0.0698)*

0.5858

(0.5139)

F-value (P-value)

11.370 (0.0001)***

26.49 (0.0001)***

49.84 (0.0001)***

Adjusted R²

0.1027

0.1973

0.2243

D-W value

2.1532

1.753

2.0790

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value.

18

Table 5: The Relationship between Diversification and Financial Performance in Taiwan’s Banking Industry (1993~2001) Dependent Variable Statistics

ROA

EPS

Net Income/Sales (NIS)

OLS Regression parameter 6.5779 (0.0001)*** -1.3007 (0.0002)***

OLS Regression parameter 2.3981 (0.0074)*** 0.2541 (0.2554)

Dummy variable for Stated-Owned and New-Private banking

-0.2217 (0.4439)

F-value (P-value) Adjusted R2 D-W value

11.14 (0.0001)*** 0.0891 1.913

Diversification SIZE

EBIT / Sales

ROE

Yuler-Walker Regression parameter 44.4155 (0.0001)*** -1.9719 (0.4456)

(PIS) Yuler-Walker Regression parameter 55.9626 (0.0001)*** -1.0951 (0.7658)

OLS Regression parameter 35.0218 (0.0005)*** 3.4620 (0.1658)

-1.0788 (0.0001)***

4.0006 (0.0963)*

1.5977 (0.6320)

-2.8023 (0.1874)

21.66 (0.0001)*** 0.1667 1.819

9.52 (0.0001)*** 0.0792 2.037

6.17 (0.0004)*** 0.0538 2.0072

7.83 (0.0001)*** 0.0620 1.739

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value.

Table 6: The Relationship between Standard Deviation of ROA and Financial Performance in Taiwan’s Banking Industry (1993~2001) Dependent Variable Statistics

ROA

EPS

Net Income/Sales (NIS)

OLS Regression parameter -0.9646 (0.0001)*** -1.6046 (0.0001)***

OLS Regression parameter -0.2913 (0.0413)*** 0.1320 (0.5826)

Dummy variable for Stated-Owned and New-Private banking

-0.3091 (0.2853)

F-value (P-value) Adjusted R2 D-W value

Sd (ROA) SIZE

EBIT / Sales

ROE

Yuler-Walker Regression parameter -9.6385 (0.0001)*** -6.0554 (0.0214)**

(PIS) OLS Regression parameter -14.0232 (0.0001)*** -8.1270 (0.0230)**

OLS Regression parameter -9.0080 (0.0001)*** -0.8286 (0.7443)

-1.1512 (0.0001)***

2.6360 (0.2546)

-1.5265 (0.5938)

-4.0602 (0.0474)**

10.245 (0.0001)*** 0.0814

20.81 (0.0001)*** 0.1600

16.389 (0.0001)*** 0.1292

15.131 (0.0001)*** 0.1193

15.863 (0.0001)*** 0.1250

1.942

1.783

2.0400

1.740

1.7570

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value. 19

Table 7: The Relationship between Capital Adequately Ratio (BIS) and Financial Performance in Taiwan’s Banking Industry (1993~2001) Dependent Variable Statistics

ROA

EPS

Net Income/Sales (NIS)

OLS Regression parameter 0.0868 (0.0001)*** 1.2756 (0.0001)***

Yuler-Walker Regression parameter 0.0596 (0.0001)*** 0.9285 (0.0003)***

Dummy variable for Stated-Owned and New-Private banking

0.1424 (0.2350)

F-value (P-value) Adjusted R2 D-W value

22.53 (0.0001)*** 0.1724 1.768

BIS SIZE

EBIT / Sales

ROE

OLS Regression parameter 0.6355 (0.0001)*** 9.5791 (0.0002)***

(PIS) OLS Regression parameter 1.5584 (0.0001)*** 18.568 (0.0001)***

Yuler-Walker Regression parameter 0.6427 (0.0002)*** 10.6979 (0.0002)***

-1.1796 (0.0001)***

3.9728 (0.0320)**

1.8543 (0.4799)

-2.2134 (0.3154)

30.24 (0.0001)*** 0.1960 2.0171

8.12 (0.0001)*** 0.0647 1.738

21.44 (0.0001)*** 0.1651 1.825

9.05 (0.0001)*** 0.0729 2.0312

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value.

Table 8: The Relationship between Market Risk (Beta) and Financial Performance in Taiwan’s Banking Industry (1993~2001) Dependent Variable Statistics

ROA

EPS

Net Income/Sales (NIS)

OLS Regression parameter

Yuler-Walker Regression parameter

Market Risk (Beta)

-0.2374 (0.1424)

SIZE

EBIT / Sales

ROE

Yuler-Walker Regression parameter

(PIS) OLS Regression parameter

OLS Regression parameter

-0.0264 (0.8887)

-4.5353 (0.0251)**

-2.6604 (0.3619)

-4.2214 (0.0535)*

0.8697 (0.0008)***

0.6717 (0.0239)**

12.0000 (0.0002)***

14.6305 (0.0018)***

8.4321 (0.0154)**

Dummy variable for Stated-Owned and New-Private banking

-0.0882 (0.6862)

-1.0160 (0.0002)***

0.8524 (0.7664)

-1.4856 (0.7060)

-4.7132 (0.1096)

F-value (P-value) Adjusted R2 D-W value

5.128 (0.0019)*** 0.0528 1.759

12.53 (0.0001)*** 0.1187 1.9991

8.0790 (0.0001)*** 0.0846 2.0616

4.2070 (0.0064)** 0.0414 1.7680

4.7040 (0.0033)*** 0.0477 1.798

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value. 20

Table 9: The Summarized Effects of the Relationship between Diversification and Risk in Taiwan’s Banking Industry (1993~2001) Independent variable

Dependent variable

Expected

Effects from

Significant

Symbol

Empirical Result

Level

Market Risk (beta) Diversification Measure (DIV)

-

Standard Deviation of ROA [sd (ROA)]

+

Capital Adequately Ratio (BIS)

-

***

+

***

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level.

Table 10: The Summarized Effects of the Relationship between Diversification and Financial Performance in Taiwan’s Banking Industry (1993~2001) Independent

Expected

Effects from

Significant

Effects

Empirical Result

Level

ROA

+

+

***

ROE

+

+

***

Measure

Net Income / Sales (NIS)

+

+

***

(DIV)

EBIT / Sales (PIS)

+

+

***

EPS

+

+

***

variable

Diversification

Dependent variable

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level.

21

Table 11: The Summarized Effects of the Relationship between Risk and Financial Performance in Taiwan’s Banking Industry (1993~2001) Independent variable

Expected Effects

Dependent variable ROA

Standard Deviation of ROA [sd(ROA)]

ROE

-

Net Income / Sales (NIS) EBIT / Sales (PIS) EPS ROA ROE

Capital Adequately Ratio (BIS)

+

Net Income / Sales (NIS) EBIT / Sales (PIS) EPS ROA ROE

Market Risk (beta)

-

Net Income / Sales (NIS) EBIT / Sales (PIS) EPS

Effects from Significant Empirical Result Level

+ + + + + -

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level.

22

*** *** *** *** *** *** *** *** *** ***

* **

Table 12: The Pairwise Granger Causality Tests of Market Risk (Beta) and Diversification in Taiwan’s Banking Industry (1993~2001) Null Hypothesis:

Lag=1

Lag=2

Obs F-value

P-value

1.141

(0.287)

Lag=3

Lag=4

Obs F-value P-value Obs F-value P-value Obs F-value P-value

Diversification does not Granger Cause Beta Beta does not

223

222 1.406

Granger Cause

2.214 (0.112)

(0.237)

1.963 (0.121) 221

1.547 (0.215)

1.404 (0.234) 220 1.995 (0.096)*

1.195 (0.313)

Diversification

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value.

Table 13: The Pairwise Granger Causality Tests of SD(ROA) and Diversification in Taiwan’s Banking Industry (1993~2001) Null Hypothesis:

Lag=1 Obs

F-value

Lag=2 P-value Obs F-value

P-value

Lag=3 Obs F-value

Lag=4

P-value Obs F-value

P-value

Diversification

28.232 (0.000)***

does not Granger Cause SD (ROA) SD (ROA) does

311

310 0.142

not Granger Cause

16.04 (0.000)***

(0.707)

11.141 (0.000)*** 309

0.276

(0.707)

9.695 (0.000)*** 308

0.551

(0.648)

0.58

(0.678)

Diversification

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value.

Table 14: The Pairwise Granger Causality Tests of BIS and Diversification in Taiwan’s Banking Industry (1993~2001) Null Hypothesis:

Lag=1 Obs

Lag=2

F-value

P-value Obs F-value

2.564

(0.11)

P-value

Lag=3 Obs F-value

Lag=4

P-value Obs F-value

P-value

Diversification does not Granger Cause BIS BIS does not Granger Cause

305

4.085 (0.018)** 304

0.05

(0.823)

2.785 (0.041)** 303

0.863

(0.423)

2.402 (0.05)** 302

0.561

(0.641)

Diversification

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value.

23

0.764

(0.549)

Table 15: The Pairwise Granger Causality Tests of ROA and Diversification in Taiwan’s Banking Industry (1993~2001) Null Hypothesis:

Lag=1 Obs

F-value

Lag=2 P-value Obs F-value

P-value

Lag=3 Obs F-value

Lag=4

P-value Obs F-value

P-value

Diversification

3.533

does not Granger Cause ROA

(0.061)*

311

ROA does not

310 1.147

Granger Cause

4.234 (0.015)**

3.128 (0.026)** 309

4.802 (0.009)***

(0.285)

2.486 (0.044)** 308

3.351 (0.019)**

3.05 (0.017)**

Diversification

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value.

Table 16: The Pairwise Granger Causality Tests of ROE and Diversification in Taiwan’s Banking Industry (1993~2001) Null Hypothesis:

Lag=1 Obs

F-value

Lag=2 P-value Obs F-value

P-value

Lag=3 Obs F-value

Lag=4

P-value Obs F-value

P-value

Diversification

8.814 (0.003)***

does not Granger Cause ROE

310

ROE does not

309 0.026

Granger Cause

8.48 (0.000)***

(0.871)

6.157 (0.000)*** 308

0.589

(0.555)

4.676 (0.001)*** 307

1.603

(0.189)

1.233

(0.297)

Diversification

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value.

Table 17: The Pairwise Granger Causality Tests of NIS and Diversification in Taiwan’s Banking Industry (1993~2001) Null Hypothesis:

Lag=1 Obs

Lag=2

F-value

P-value Obs F-value

1.865

(0.173)

P-value

Lag=3 Obs F-value

Lag=4

P-value Obs F-value

P-value

Diversification does not Granger Cause NIS NIS does not Granger Cause

309

2.987 (0.052)* 307

0.08

(0.778)

2.262 (0.081)* 305

0.273

(0.761)

2.118 (0.079)* 303

0.543

(0.653)

Diversification

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value.

24

0.697

(0.595)

Table 18: The Pairwise Granger Causality Tests of PIS and Diversification in Taiwan’s Banking Industry (1993~2001) Null Hypothesis:

Lag=1 Obs

Lag=2

F-value

P-value Obs F-value

1.819

(0.178)

P-value

Lag=3 Obs F-value

Lag=4

P-value Obs F-value

P-value

Diversification does not Granger Cause PIS

311

PIS does not

310 0.498

Granger Cause

4.829 (0.009)***

(0.481)

3.366 (0.019)** 309

1.677

(0.189)

3.057 (0.017)** 308

1.013

(0.387)

1.3

(0.272)

Diversification

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value.

Table 19: The Pairwise Granger Causality Tests of EPS and Diversification in Taiwan’s Banking Industry (1993~2001) Null Hypothesis:

Lag=1 Obs

F-value

Lag=2 P-value Obs F-value

P-value

Lag=3 Obs F-value

Lag=4

P-value Obs F-value

P-value

Diversification

8.744 (0.003)***

does not Granger Cause EPS EPS does not Granger Cause

310

5.39 (0.005)*** 309

1.422

(0.234)

5.628 (0.001)*** 308

4.752 (0.009)***

4.681 (0.017)** 307

5.342 (0.001)***

Diversification

*Significant at 10% level. ** Significant at 5% level. *** Significant at 1% level. Note: Parenthesis shows the P-value.

25

4.064 (0.003)***

Reference 1. Gordon V. Karels, M. Kabir Hassan and Manfred O. Peterson (1994). Deposit insurance, market discipline and off-balance sheet banking risk of large U.S. commercial banks, Journal of Banking and Finance, 18, pp.575-593.

2. Ghemawat, P. (1991), Commitment. New York: Free Press. 3. Granger, C. W.J. (1969). Investigating Causal Relations by Econometric Models and Cross-Spectral Method, Econometrica, 52, 701-720.

4. Kevin Rogers and Joseph F. Sinkey Jr. (1999). An analysis of nontraditional activities at U.S. commercial banks, Review of Financial Economics, 8, pp.25-39.

5. M. Kabir Hassan (1993). The off-balance sheet banking risk of U.S. commercial banks, The Quarterly Review of Economics and Finance, 33 (1), pp.51-69.

6. Montgomery, C.A. (1994). Corporate diversification, Journal of Economic Perspectives, 8 (67), May, pp.163-178.

7. Rebecca S. Demsetz and Philip E.Strahan (1997). Diversification, size, and risk at bank holding companies, Journal of Money, Credit, and Banking, 29 (3), August.

8. Rebecca S. Demsetz and Philip E.Strahan (1995). Historical patterns and recent changes in the relationship between bank holding company size and risk, Economic Policy Review – Federal Reserve Bank of NEW YORK, 1 (2), July.

9. Wernerfelt, B. (1984). A resource-based view of the firm, Strategic Management Journal, 5, 171-180.

26