Evaluate the impact of BDIV on the performance and risk of firms; assess the impact of ADIV on the performance and risk of firms through the mediating effect of BDIV; evaluate the impact of the CDIV on the performance and risk of firms through the mediating effect of ADIV.
MINISTRY OF EDUCATION AND TRAINING HO CHI MINH CITY OPEN UNIVERSITY VU HUU THANH THE IMPACT OF DIVERSIFICATION ON PERFORMANCE AND RISK OF FIRMS Major : Business Administration Major Code : 62 34 01 02 SUMMARY OF THE PHD THESIS HCMC, 2018 The thesis is completed at the: HO CHI MINH CITY OPEN UNIVERSITY Reviewer: 1:……………………………………………………………………… Reviewer: 2:……………………………………………………………………… The thesis will be defended in front of the Thesis Committee at Academy level, at the Open University of Ho Chi Minh city At…… date …… month …… year 2018 CHAPTER INTRODUCTION 1.1 Research problem Diversification is considered as one of the strategic activities of a firm (Ramanujam and Varadarajan, 1989) and it has a profound effect on many aspects of firms; notably, its impact on performance and risk Following the structure-conduct-performance paradigm (S – C – P paradigm) in theory of industrial organization (IO theory), the output of a firm is linearly related to the market structure (Weiss, 1979) The S – C – P paradigm describes the structure of the market determines the behaviour of the firms, and then the behaviour of the firms determines their output (McWilliams and Smart, 1993) The S – C – P framework is integrated into the theory of strategic management (Jemison, 1981) to describe the impact of a variety of strategic activities, including diversification, on business performance (McWilliams and Smart, 1993) Thus, derived from industrial organization theory, diversification is regarded as a component of strategic management and this component has significant implications for firms’ output Besides performance, risk also receives particular attention in strategic management The link between diversification and risk was soon considered by Bettis and Hall (1982) and has since formed a direction to this day The impact of diversification on risk is quite complicated This complexity is due to the diversity of the type of firm diversification, each of which diversifies into lower-level diversification based on different classification Different types of diversification at the lower level tended to have different impacts on risk, and each type of diversification itself may have certain relationships For example, business diversification in a firm is divided into two broad categories which are related and unrelated diversification Diversification of related business can increase performance (Palich et al., 2000) but also increase specific risk (Bettis and Hall, 1982) Meanwhile, unrelated diversification has the opposite effect Based on the theoretical and empirical studies on the effects of diversification on the performance and risk, the thesis found some following issues: Firstly, the study of the impact of business diversification (BDIV) is prevalent in the world However, the dissertation has not found any similar study up to 2015 By 2016, the research by Santarelli and Tran (2016) on the effect of diversification on the performance of Vietnamese enterprises appears The study used GSO data to measure overall BDIV and to estimate the impact of this variable on business performance The study has not conducted an estimation of the impact of various types of diversification on performance; in particular, the effect of related and unrelated business on performance According to Palich et al (2000) and Benito-Osorio et al (2012), the related and unrelated BDIV tended to have different effects on the performance due to the distinctive characteristics of the two types of diversification Thus, by combining these two types of diversification into overall business diversification, it is difficult for firms to choose which type of diversification is appropriate for their business In practice, firms need to decide what type of diversification to pursue (diversification of the related business, diversification of the unrelated business, geographic diversification, international diversification, ) rather than pursue diversification in general Secondly, the impact of BDIV on risks, especially bankruptcy risk, which has rarely been studied in the empirical literature, has been reported in about 17 studies up to 2016 Bankruptcy risk variables are primarily measured by the Atman Z - Score, a measure that is only suitable for the US market that may not be suitable for an emerging market like Vietnam Thirdly, the study of the effects of BDIV on the performance and bankruptcy risk is often conducted separately in individual research models without being simultaneously performed in an empirical model In addressing the three issues mentioned above, the thesis deals with the first topic titled "The impact of business diversification on performance and risk." This topic will explore the impact of related and unrelated business diversification on performance, and this research employs the Linear Structural Model (SEM) to estimate these relationships Also, firm performance variables will be viewed from a shareholder's perspective and measured by ROE (return on equity), the bankruptcy risk variable is measured MKV model (probability of default) Using SEM model and MKV bankruptcy risk measurement can be considered as the new point of the research Fourth, diversification can be seen in various angles when placed in relation to the performance and risk of a firm Diversification can be seen in terms of business diversification (Asoft, 1957), product diversification (Rumelt, 1974), geographic diversification (Stopford and Wells, 1972; Daniels and Bracker, 1989; and Sullivan, 1994), merger diversification (Morck et al., 1990), and customer diversification (Hsu and Liu, 2008) These types of diversification are commonly referred to as business diversification because they are all oriented towards the business of firms However, there are other forms of diversification taking place in the internal environment of a firm such as capital diversification or asset investment diversification have been not be explored The three types of diversification are capital diversification (CDIV), asset investment diversification (ADIV), and BDIV that may have a specific relationship with each other in the context of corporate financial decision-making In particular, CDIV may impact on ADIV and then ADIV may impact on BDIV In addition, ADIV and CDIV may also affect performance and risk if based on such theories as Rumelt's "escape hypothesis" (Rumelt, 1974), theory of resource-based view” (Wernerfeld, 1984 and Barney,1991), the capital structure theory (Modilligani and Miler, 1958 and Myers, 1977), and theory of free cash flow of Jensen (1986) Given the above, two questions are posed: first, is there a practical effect of ADIV on performance and the risk of bankruptcy through the mediating effect of BDIV? Moreover, secondly, is there a practical effect of the CDIV on performance and risk through mediating effect of ADIV? In response to these two questions, the thesis continues to carry out two further research topics The second research topic titled "The impact of asset investment diversification on business diversification, performance and risk," and the third research topic titled "The impact of capital diversification on asset investment diversification, performance and risk " On the second research topic, the dissertation will classify the ADIV into two components, namely, related assets and unrelated assets to the core business of a firm Then the dissertation builds the theoretical relationship among ADIV, BIDV, performance, and risk Then, the thesis builds the SEM model to estimate these relationships empirically Building the links among ADIV, BDIV, performance, and risk and employing SEM model to estimate these links can be considered as a new point in the thesis Similarly, in the third research topic, the thesis will classify CDIV into equity diversification and debt diversification The thesis then builds the theoretical relationship among CDIV, ADIV, performance and risk Next, the thesis employs the SEM model for empirical estimation Establishing the links among CDIV, ADIV, performance, and risk, as well as estimating with the SEM model, may be considered as a new point in the thesis 1.2 Research questions The thesis should answer three research questions as follows: Question 1: How does the BDIV affect performance and risk? Question 2: How does ADIV affect performance and risk through the mediating effect of BDIV? Question 3: How does the CDIV affect performance and risk through the mediating effect of ADIV? 1.3 Research objectives The dissertation should address three research objectives as follows: Objective 1: Evaluate the impact of BDIV on the performance and risk of firms Objective 2: Assess the impact of ADIV on the performance and risk of firms through the mediating effect of BDIV Objective 3: Evaluate the impact of the CDIV on the performance and risk of firms through the mediating effect of ADIV 1.4 Object and scope of the study The object of the study is the impact of the diversification on the performance and risk of a firm More specifically, the impact of BDIV on the performance and risk of firms; the impact of ADIV on the performance and risk of firms through mediating effect of BDIV; and impact of the CDIV on the performance and risk of firms through the mediating effect of ADIV The research scope is non-financial businesses listed on the Ho Chi Minh City Stock Exchange and the Hanoi Stock Exchange between 2008 and 2015 1.5 Research contents Based on research, the research questions and objectives, the dissertation has four main research contents as follows: Firstly, a theoretical study on the linkages between the three types of diversification, including the BDIV, ADIV, and CDIV Second, a theoretical study on the linkage among the three types of diversification, performance, and risk Third, develop a research method to empirically estimate relationships mentioned above, including (i) BDIV, performance, and risk; (ii) ADIV, BDIV, performance, and risk; (iii) CDIV, ADIV, performance, and risk Fourth, empirically estimate these relationships and discuss results 1.6 New research contributions The thesis has three research topics to address three research objectives and has new contributions described as followings: Topic one: The impact of business diversification on performance and risk This topic is performed in chapter two The research in topic one is still a repetitive study of the impact of BDIV on performance and risk However, by measuring the risk of bankruptcy using the Merton-KMV model (probability of default model), and estimated these relationships using the SEM model, the research in the topic has contributed in terms of: (i) measuring the risk of bankruptcy by Merton-KMV model and (ii) employing SEM model to estimate empirically Topic two: The impact of ADIV on BDIV, performance and risk The research is addressed in chapter three The second topic considered ADIV as related and unrelated assets diversification and built the relationship between these two variables with BDIV, performance, and risk Thus, the second research topic has contributed regarding: (i) create new concepts (related and unrelated asset investment diversification) and (ii) build new link among concepts (ADIV, BDIV, performance, and risk) Topic three: The impact of CDIV on ADIV, performance, and risk The topic is organized in chapter four The research on the topic focused on understanding the CDIV The study analyzes CDIV as two components: equity diversification and debt diversification Then the thesis establishes the linkage of these two variables to the ADIV, performance, and risk Thus, similar to topic two, topic three has contributed to the following new points: (i) build new concepts (equity diversification and debt diversification) and (ii) establish new relationships among concepts (CDIV, ADIV, performance, and risk) 1.7 Thesis structure The thesis is organized into five chapters with the following title: Chapter Introduction Chapter Impact of business diversification on performance and risk Chapter Impact of Asset investment diversification on business diversification, performance, and risk Chapter The Impact of capital diversification on asset investment diversification, performance, and risk CHAPTER THE IMPACT OF BUSINESS DIVERSIFICATION ON PERFORMANCE AND RISK 2.1 Literature review 2.1.1 ĐDH This section explores four issues: (i) concept of diversification, (ii) classification of diversification, (iii) identification of diversification content used in the thesis, and (iv) measure diversification 2.1.1.1 Khái niệm ĐDH Based on previous studies, the thesis presents the concept of two types of business diversification as follows: Related business diversification (RBDIV) occurs when a firm adds a new line of products related to existing core product Unrelated business diversification occurs when a firm expands a new line of products unrelated to existing core product 2.1.1.2 Diversification measurement The favourite way to measure diversification in studies is to combine the classification of Rumelt (1974) and the entropy method The first author proposed using entropy to calculate the degree of diversification was Jacquemin and Berry (1979) Then Palepu (1985) based on Rumelt's classification and associated with the theory of entropy calculated a firm diversification, including total diversification, related diversification, and unrelated diversification Whereby: Total diversification: 𝑡 𝐷𝑇 = − ∑ 𝑝𝑡 log ( ) 𝑝𝑡 𝑡=1 Of which: “t” is the tth segment, “pt” is the sales of the tth segment to total sales Related diversification: 𝑟 𝐷𝑅 = ∑ 𝑝𝑟 log ( ) 𝑝𝑟 𝑟=1 Of which: r is the rth segment related to the primary business, “pr” is the sales of the rth segment to total sales Unrelated diversification: 𝑢 𝐷𝑈 = ∑ 𝑝𝑢 log ( ) 𝑝𝑢 𝑢=1 Of which: “u” is is the uth segment unrelated to the primary business, “pu” is the sales of the rth segment to total sales 2.1.2 Performance of Firms 2.1.2.1 Definition This study approaches the concept of "business performance " from a shareholder's point of view, so corresponding to this angle is the profit aspect of firms Profit is considered the common factor used in most studies on the effects of diversification on performance 2.1.2.2 Performance measurement This research measures performance in a shareholder's perspective because shareholders will be interested in the company's strategic activities, including diversification Also, this is also a measure of the accounting performance of firms and is widely used by many studies (Lee and Li, 2012) The variable employed to measure performance under shareholder’s perspective is "return on equity" (ROE) 2.1.3 Bankruptcy risk 2.1.3.1 Definition Bankruptcy risk is considered as "the probability of occurrence of an event where a firm is unable to pay its debts when it is due" 2.1.3.2 Measurement This research approaches the risk from the perspective of the probability of bankruptcy of Merton (1974), Kealhofer and Bohn (1998), McQuown (1997) and Vasicek (1984), which is named Merton-MKV model We employ this model to estimate the probability of bankruptcy of a firm 2.1.4 The relationship between BDIV and performance 2.1.4.1 Theories on the effect of BDIV on performance Theory of firm growth According to Penrose (1959), the goal of business owners is to increase long-term profitability and to achieve this goal the firm must achieve growth through expansion including diversification This theory indirectly explains the relationship between BDIV and performance firm growth Economies of scope The theory of economies of scope was developed by Panzar and Willig (1977) This theory states that the average cost of production falls when firms expand on the type of goods and services it produces Firms get a cost advantage when they produce complementary products while they still concentrate on their core competencies As firms diversify their products, especially the diversification of related products, they will have the opportunity to convert resources and share resources, such as skills and technology… (Rumelt, 1982; Teece, 1982; Markides and Williamson, 1994; and Barney, 1997) If two or more businesses of firms share some common resources with each other, the total cost of production will decrease; hence, the efficiency will increase This theory is used to explain the positive relationship between RBID and performance Theory of resource-based view – RBV Wernerfelt (1984) and Barney (1991) then developed Penrose's perspective on RBV theory According to Barney (1991), firms’ resources including all kinds of assets, capabilities, processes, knowledge, are controlled by firms and can help firms perform effective strategies Resources will provide a sustainable competitive advantage if they have characteristics that are: valuable, rare, imperfectly imitable, strategically nonsubstitutable This theory is used to explain the negative impact of UBIDV on performance 2.1.4.2 Empirical evidence on the impact of BDIV on performance The thesis explored 71 articles and 51 articles on the impact of the RDIV and UDIV on performance alternately The study found that for the RDIV variable, the positive impact dimension was dominant (45.6%); In contrast, negative influences account for a majority (39.2%) in the studies the effect of UDIV on performance Based on the theory, empirical studies and meta-analyses, the thesis hypothesizes the relationship between BDIV and performance as follows: Hypothesis H1.1: RBDIV tends to have a positive impact on the performance of a firm Hypothesis H1.2: UBDIV tends to affect the performance of a firm negatively 2.1.5 BDIV and risk 2.1.5.1 Theories on the effect of BDIV on performance Lubatkin Chatterjee (1994a) Lubatkin and Chatterjee (1994) research the impact of business diversification on the specific and systematic risk of a firm They argue that portfolio theory may not be applied to all cases of the impact of business diversification on risk; notably, the impact of business diversification on the specific risk (unsystematic risk) They conclude that related diversification reduces unsystematic risk; conversely, unrelated diversification has a positive effect on this risk Their conclusion is contrary to the theory of portfolio They explain that firms diversifying into related industries own some synergistic interrelationship (tangible and intangible interrelationship), so that specific risk is reduced On the contrary, when firms expand their business into unrelated industries, they will lack internal relationships, and it will lead to the increase in unsystematic risk Bettis Hall (1982) Bettis and Hall (1982) study on the relationship between business diversification and accounting risk They state that related diversification has a lower effect on accounting risk than unrelated diversification They explain as follows: Assuming that a firm operates only in one industry and it gains profit R1 Based on the theory of tradeoffs between profit and risk, the variation of R1 (Var (R1)) is the accounting risk of the firm The meaning of the Var (R1) is that if the profit of firm increases or decreases considerably in successive periods of time, the risk rises rapidly Next, the firm expands from the existing industry to other R1 and R2 alternatively represent the profit of the first and the second industry, the combined risk of both industries gets measured by the sum of the variances of both R1 and R2: Var(R1 + R2) = Var(R1) + 2Cov(R1,R2) + Var(R2) If Cov(R1, R2) is positive or the correlation between returns is positive, the combined risk is higher Conversely, if the correlation is negative, the firm gets the smaller sum of risk Bettis and Hall (1982) suggest that when a firm operates in two related industry, Cov(R1, R2) is positive On the contrary, if a firm operates in two unrelated industry, Cov(R1, R2) is negative These imply that the firm copes with the problem of the business cycle when they enter more than an industry Unrelated diversification can eliminate the cycle and lower the risk while related diversification can exacerbate firm business cycle downturn and raise the combined risk From here, the thesis presents the relationship between business and bankruptcy risk as follows: Hypothesis H1.3: The risk of bankruptcy of a firm can increase when the firm conducts RBDIV Hypothesis H1.4: The risk of bankruptcy of a firm can be reduced if the firm conducts UBDIV 2.1.5.2 Empirical evidence on the impact of BDIV on risk The following table summarizes the impact of BDIV on specific risks Table 2.12 Summarize the impact of business diversification to specific risks Direction Possitive effect Negative effect No effect Tổng RBDIV Frequency 17 UBDIV Percent 29.4% 52.9% 17.7% 100% Frequency 10 17 Percent 58.8% 29.4% 11.8% 100% 18 3.2.2.2 BDIV and ADIV measurement Measurement of BDIV was described in topic one ADIV is measured as follows: (i) Related assets: 𝑛 𝑅𝐴_𝐷𝐼𝑉 = ∑ 𝑝𝑟 log ( ) 𝑝𝑟 𝑟=1 (i) Unrelated assets: 𝑛 𝑈𝐴_𝐷𝐼𝑉 = ∑ 𝑝𝑢 log ( ) 𝑝𝑢 𝑢=1 3.2.2.3 Control variable Control variables affecting the two output variables (performance and risk) were identified in topic one 2.3.2.3 Estimation With the proposed theoretical and empirical model of research, the study used RB_DIV and UB_DIV as two mediators So this model is also known as the mediation model or path model which is a kind of SEM (Stata, 2015) 3.2.4 Data Described in Chapter 3.3 Analysis and result discussion This analysis will begin with statistical analysis describing variables as well as regression The study will then conduct a discussion based on the regression results 3.3.1 Descriptive statistics 3.3.2 Regression results 3.3.2.1 Regression results and SEM model tests This study implements SEM regression analysis; then we perform the necessary tests such as (i) good-of-fitness test, (ii) homoscedasticity test, and (iii) normal distribution of the residual test The thesis finds that SEM fails to satisfy homoscedasticity test and normal distribution of the residual test Hence, this research will employ GSEM to replace SEM 3.3.2.2 GSEM estimation results As mentioned above, SEM model regression has encountered two problems; then the regression results are not biased but inconsistent We propose GSEM model combined with 19 robust estimation techniques to replace the original model The regression coefficients of the GSEM model are not different from the ones of SEM model but more robust than SEM model The results of GSEM model described below: Figure 3.6 Summary of GSEM estimation Notation Name of variables Affecting ROE RB_DIV Related business Expectation Coef Robust Std Err P>z + 0.095471 0.0028 0.000 UB_DIV Unrelated business - -0.122959 0.0012 0.000 RA_DIV Related assets + 0.025516 0.0079 0.001 UA_DIV Unrelated assets - 0.016931 0.0130 0.193 AGE Firm age + -0.000849 0.0009 0.351 SG Sales growth + 0.000162 0.0004 0.700 CR Current payment ratio 0.000197 0.0019 0.918 -0.033490 0.0120 0.005 0.004662 0.0007 0.000 CCC The ratio of operating costs to revenue Natural log of total assets + Cash conversion cycle - -0.000001 0.0000 0.596 AS Assets structure - -0.160113 0.0259 0.000 AT Assets turnover + 0.040105 0.0076 0.000 SGA_S LNA Affecting RISK RB_DIV Related business Unrelated business + 0.006920 0.0019 0.000 UB_DIV - -0.003736 0.0009 0.000 RA_DIV Related assets + 0.010889 0.0055 0.050 UA_DIV Unrelated assets + 0.061860 0.0091 0.000 CR Current payment ratio -0.000997 0.0013 0.453 0.026354 0.0083 0.002 0.000000 0.0000 0.259 P_B The ratio of operating costs to + revenue Interest Coverage Ratio Price to book ratio - 0.010485 0.0051 0.038 LNA Natural log of total assets - 0.000112 0.0005 0.825 CCC Cash conversion cycle + 0.000000 0.0000 0.730 OCF_S Operating cash flow ratio Assets turnover - -0.010195 0.0049 0.038 - -0.005265 0.0052 0.314 + -0.036146 0.1720 0.834 SGA_S EBIT_I AT Affecting UB_DIV UA_DIV Unrelated assets Tác động tới RB_DIV 20 RA_DIV Related assets + 0.091767 0.0470 0.051 3.3.2.3 Discussion Based on the results of the regression, the thesis discusses the effect of ADIV on BDIV UB_DIV, ROE, and RISK as follows: First, regarding structural impact, the RA_DIV variable has a direct and indirect effect on ROE and RISK In this case, the RB_DIV has become the mediating variable that transmits the impact of RA_DIV on ROE and RISK In contrast, UA_DIV does not show the structural impact to both ROE and RISK because UB_DIV does not become the mediator transmitting UA_DIV's impact on the ROE and RISK Second, the effect of RA_DIV on ROE and RISK is a trade-off RA_DIV has a positive impact on ROE, but in return is RISK However, this tradeoff is not significant when the coefficient of RA_DIV on RISK is quite low (1.063%) In contrast, the impact of UA_DIV on ROE and RISK shows no a trade-off UA_DIV has a positive effect on RISK, but this does not show any statistically significant effects on ROE Third, regarding the overall impact on ROE The magnitude impact of RA_DIV is greater than that of UA_DIV (3.37% vs 0.00%) Fourth, in terms of total marginal impact on RISK The magnitude of the impact of RA_DIV is higher than that of UA_DIV (1.063% vs 0.62%) Final, related to marginal impact level, RA_DIV has a positive effect on RB_DIV whereas UA_DIV does not show any effect on UB_DIV 3.4 Conclusion and recommendation 3.4.1 Conclusion Firstly, the SEM estimation model in the thesis satisfies most of the required tests except for two tests that are heteroskedasticity test and normal distribution of residuals test From here, the study recommends employing GSEM model for estimation Secondly, regarding the structural impact of RA_DIV on ROE, the thesis explores the direct impact of RA_DIV on ROE and the indirect impact of RA_DIV on ROE through RB_DIV RA_DIV has a statistically significant direct impact on ROE with a marginal impact of 0.025 Thirdly, the thesis also evaluates the structural effect of UA_DIV on ROE Regarding direct impact, the coefficient of UA_DIV is not statistically significant, which means that there is no direct impact on the model From the angle of indirect impact, the effect of UA_DIV on UB_DIV is also not statistically significant Thus, the total impact of UA_DIV on ROE is zero The thesis also investigates the structural effect of UA_DIV on RISK, in which the direct effect is positive and statistically significant (equal to 0.062), but the indirect effect level is zero As a result, the total impact of UA_DIV on RISK is equal to the level of direct impact 21 Finally, the impact of RA_DIV on ROE and RISK is on a trade-off ROE has a positive impact on ROE, but RISK may be increased 3.4.2 Recommendation Based on the results of the analysis, the thesis gives some recommendations related to the business activities of firms as follows: First, to obtain better returns in the short term and at the same risk of bankruptcy, firms should focus on RADIV Increasing RA_DIV also increases RB_DIV then increase ROE Reducing UADIV does not reduce the ROE in the short run but reduces RISK Second, in the long run, the impact of the investment may change the structural impact to ROE and RISK, especially the impact of UA_DIV Delay effect may occur in the model Therefore, firms need to monitor the historical data of investment to decide whether to expand or narrow each type of investment 3.4.3 Limitations and further research 22 CHAPTER4 THE IMPACT OF CAPITAL DIVERSIFICATION ON ASSET INVESTMENT DIVERSIFICATION, PERFORMANCE AND RISK 4.1 Literature review 4.1.1 Capital diversification definition and measurement 4.1.1.1 Capital diversification definition Owner diversification (ODIV): a firm uses its equity capital with different characteristics to form the equity Debt diversification (DDIV): a firm use different types of debt financing to form the total debt 4.1.1.2 Capital diversification measurement Owner diversification 𝑛 𝑂 𝐷𝑖𝑣 = ∑ 𝑝𝑜 ln( ) 𝑝𝑜 𝑜=1 Debt diversification: 𝑛 𝐷_𝐷𝐼𝑉 = ∑ 𝑝𝑙 ln( ) 𝑝𝑙 𝑙=1 4.1.2 The link between CDIV and ADIV As mentioned, the thesis divides the capital diversification into two main components: owner and debt diversification It is difficult to determine which source of capital will finance what kind of assets However, based on research by Montgomery and Singh (1984), Rajagopalan and Harrigan (1986), Lubatkin and O'Neill (1987), and Barton (1988), unrelated business activities are risky therefore they are financed by equity On the other hand, related business activities are less risky, so they are likely to be financed by debt Also, according to Rauh and Sufi (2010), in all sources of capital, equity is the most sensitive information Because, based on the theory of agency, the agents easy to use equity to perform overinvestment for the sake of themselves, or address risky investment Meanwhile, the use of debt financing for risky projects is difficult because creditors tightly control it From here, the thesis gives two research hypotheses: Hypothesis H3.1: Debt diversification may increase RADIV Hypothesis H3.2: Owner diversification will have a positive effect on UADIV However, the level of its impacts will be lower than the impact of debt diversification 4.1.3 The links among CDIV, performance and risk 4.1.3.1 ODIV, performance and risk 23 Agency theory According to Jensen and Meckling (1976), the agent tends to allocate the resources of a firm to serve their interests Diversification is also a means of allocating resources of representatives Through this activity, the agent wants to consolidate the position or satisfy the personal interest To address the "Agency problem I," Amihud and Lev (1981) suggest that the firm needs to be financed by a large shareholder The owner has more power to reject additional proposals through diversification Through empirical research (Amihud and Lev, 1981), it is found that a firm with concentrated ownership generally decrease diversification There are some issues to consider as follows: Firstly, the more dispersed shareholders, the more seriously "Agency problem I" will appear; Secondly, to solve the "Agency problem I," the firm needs to have a concentrated ownership structure However, the concentrated ownership has a twofold effect There are different types of shareholders; each has a different motivation Therefore when the company focuses on each type of shareholder, it will benefit from the characteristics of the shareholder but will also receive disadvantage of this type Finally, in the context of diversification, firms have different sources of equity, it means that firms disperse the right to make decisions It will be difficult for owners to prevent representatives’ decisions on diversification This lead to the performance may decline, but the risk of bankruptcy will be reduced since firms own various financing source Research hypothesis about the impact of CDIV on performance and risk: Hypothesis H3.3: ODIV reduces the performance of the firm Hypothesis H3.4: The bankruptcy risk of the business will decrease as the firm expands its equity financing 4.1.3.2 The links among DDIV, performance, and risk (i) Free cash flow theory and capital structure theory Free cash flow theory Conflicts of interest between shareholders and representatives increase as firms have free cash flow (Jensen, 1986) When free cash flow occurs, the agency costs of free cash flows also appear It means that managers prefer using free cash flow to invest in projects for personal interest although these projects are less effective for firms The problem is how to prevent managers from using this cash flow to invest in inefficient projects while still ensuring firm growth This is a problematic contradiction because using the entire free cash flow to pay dividends leads to the firm having less money to invest and losing growth momentum in the future, resulting in reducing dividends in the future and the stock price will fall Meanwhile, 24 managers still have to make a promise to pay dividends with a suitable growth in the future Failing this promise, the manager will face many disadvantages in terms of position or salary To resolve this conflict of interest, a firm may decide to increase its debt Increasing debt binds the manager to the promise of paying future cash flow to creditors in addition to paying the owner's cash flow (via dividends) Capital structure theory The theory of capital structure was developed by Modilligani and Miler (1958) and used by empirical studies to explain the relationship of capital structure to performance According to Modilligani and Miler (1958), capital structure is a combination of short-term debt, long-term debt and equity This amount will be used to finance various investment A company can have an optimal capital structure by using appropriate financial leverage Thus, businesses can lower their cost of capital by increasing the use of debt (because when using corporate debt they will benefit from the tax shield) However, when the rate of debt increases, firms face two risks: (i) the liquidity risk increases and (ii) the risk to lose equity The combination of two risks will leave the benefits of debt and lead the firm to financial distress circumstance (Brennan and Schwartz, 1978) In addition, according to Trade-off Theory (Myers, 1977), debt financing has the advantage of benefiting from the tax shield At medium levels of debt, the probability of financial exhaustion is negligible, and the present value of financial exhaustion is small so that debt financing will be an advantage for firms However, as a firm get more and more debt, the risk of bankruptcy will be higher and then lower the value of the business At some point, the value of the tax shield is equal to the value of the financial exhaustion Debt structure issues According to Rauh and Sufi (2010), different sources of debt have different effects on firms because the information hidden within each type of debt is differently sensitive According to Rauh and Sufi (2010), bank debt contains the least sensitive information, as these are usually collateralized The type of debt which has the most sensitive information is a convertible bond Hence, aggregating debt from different types has led to the various impact of total debt on performance It is difficult to explain the harmful effects of debt on performance To solve this problem, it is necessary to classify debt into different types based on the level of information sensitivity Rauh and Sufi (2010), based on the classification of the US Securities and Exchange Commission, have divided the company's debt into seven categories: (i) Bank debt, (ii) Bond, (iii) Convertible debt, (iv) Private placement, (v) Mortgage or equipment debt, (iv) Program debt such as corporate or medium-term notes, and (vii) Other liabilities Hypothesis on the impact of DDIV on performance and risk Hypothesis H3.5: The performance of a firm will increase higher when the firm conduct debt financing 25 Hypothesis H3.6: The risk of bankruptcy will increase as firms expand their debt financing 4.1.4 The links among CDIV, ADIV, performance and risk Hình 4.4 The links among CDIV, ADIV, performance and risk 4.2 Methods 4.2.1 Empirical research model Figure 4.5 Empirical research model 4.2.2 Variable measurement 4.2.2.1 Performance measurement The measurement was described in topic one 4.2.2.2 Risk measurement The measurement was described in topic one 26 4.2.2.3 Đo lường biến số ĐDH đầu tư tài sản The measurement was described in topic two 4.2.2.4 CDIV measurement (i) ODVI 𝑛 𝑂_𝐷𝐼𝑉 = ∑ 𝑝𝑜 log ( ) 𝑝𝑜 𝑜=1 (ii) DDIV 𝑛 𝐷_𝐷𝐼𝑉 = ∑ 𝑝𝑙 log ( ) 𝑝𝑙 𝑙=1 4.2.2.5 Control variable Control variables affecting the two output variables (performance and risk) were identified in topic one 4.2.3 Estimation With the proposed theoretical and empirical model of research, the study used RA_DIV and UA_DIV as two mediators So this model is also known as the mediation model or path model which is a kind of SEM (Stata, 2015) 4.2.4 Dữ liệu nghiên cứu Described in Chapter 4.3 Analysis and result discussion 4.3.1 Descriptive statistics 4.3.2 Regression results 4.3.2.1 Regression results and SEM model tests This study implements SEM regression analysis; then we perform the necessary tests such as (i) good-of-fitness test, (ii) homoscedasticity test, and (iii) normal distribution of the residual test The thesis finds that SEM fails to satisfy normal distribution of the residual test Hence, this research will employ GSEM to replace SEM 4.3.2.2 GSEM estimation results As mentioned above, SEM model regression has encountered the problems; then the regression results are not biased but inconsistent We propose GSEM model combined with robust estimation techniques to replace the original model The regression coefficients of the 27 GSEM model are not different from the ones of SEM model but more robust than SEM model The results of GSEM model described below: Figure 4.7 Summary of GSEM estimation Notation Name of variables Affecting ROE UA_DIV Unrelated assets Expectation Coef Robust Std Err P>z + 0.01449 0.0259 0.575 RA_DIV Related assets - 0.02710 0.0158 0.086 O_DIV Owner diversification - -0.02065 0.0274 0.452 D_DIV Debt diversification - 0.09285 0.0150 0.000 AGE Firm age + -0.00260 0.0018 0.149 SG Sales growth + 0.00016 0.0008 0.843 CR Current payment ratio - -0.00042 0.0038 0.910 -0.03955 0.0237 0.096 0.00516 0.0014 0.000 CCC The ratio of operating costs to revenue Natural log of total assets + Cash conversion cycle - 0.892 AS Assets structure - -5.30E- 0.0000 07 -0.08800 0.0515 AT Assets turnover + 0.03392 0.0152 0.025 Affecting RISK UA_DIV Unrelated assets SGA_S LNA 0.088 + 0.06265 0.0091 0.000 RA_DIV Related assets - 0.00995 0.0056 0.074 O_DIV Owner diversification + -0.02455 0.0097 0.011 D_DIV Debt diversification + 0.02007 0.0054 0.000 CR Current payment ratio -0.00106 0.0013 0.423 0.02551 0.0084 0.002 8.31E-08 0.0000 0.300 P_B The ratio of operating costs to + revenue Interest Coverage Ratio Price to book ratio - 0.01392 0.0051 0.007 LNA Natural log of total assets - 0.00011 0.0005 0.826 CCC Cash conversion cycle + 5.91E-07 0.0000 0.665 OCF_S Operating cash flow ratio Assets turnover - -0.01053 0.0049 0.032 - -0.00684 0.0053 0.193 SGA_S EBIT_I AT Affecting UA_DIV 28 O_DIV Owner diversification Affecting RA_DIV Debt diversification D_DIV + 0.05259 0.0172 0.002 + 0.06318 0.0219 0.000 4.3.2.3 Discussion Based on the results of the regression, the thesis discusses the effect of CDIV on ADIV, ROE, and RISK as follows: First, regarding structural impact, the D_DIV variable has a direct and indirect impact on ROE and RISK In this case, the variable RA_DIV has become the mediating variable that transmits the impact of D_DIV on ROE and RISK Meanwhile, O_DIV did not show the structural impact on ROE but showed the structural impact on RISK As a consequence, the UA_DIV variable becomes mediator affecting RISK but does not become the mediator affecting ROE Second, the impact of D_DIV on ROE and RISK is on a trade-off D_DIV has a positive impact on ROE, but in return is RISK However, this tradeoff is not significant when the level of impact of D_DIV on RISK is quite low (2.063%) In contrast, the impact of O_DIV on ROE and RISK shows no a trade-off O_DIV has a negative impact on RISK but does not show any statistically significant effect on ROE Third, regarding the degree of impact on ROE, the magnitude of the impact of D_DIV is greater than O_DIV (9.5% vs 0.00%) Fourth, in terms of marginal impact level on RISK, the margin of effect of D_DIV is higher than that of UA_DIV (2.063% vs -2.20%) Fifth, regarding the marginal impact level on the mediator variable, both O_DIV and D_DIV have positive effects on the mediator variables UA_DIV and RA_DIV Final, the marginal impact level of D_DIV to RA_DIV is higher than the marginal impact of O_DIV to UA_DIV (0.0632 vs 0.0526) 4.4 Conclusion and recommendation 4.4.1 Conclusion Firstly, the SEM estimation model in the thesis satisfies most of the required tests except for normal distribution of residuals test From here, the study recommends employing GSEM model for estimation Secondly, regarding the structural impact of O_DIV to ROE, the thesis explores the direct impact of O_DIV on ROE and the indirect impact of O_DIV on ROE through UA_DIV O_DIV does not show any direct impact on ROE In addition, regarding indirect effects, the magnitude of the impact of UA_DIV on ROE is 0.00, so the overall impact level of O_DIV on ROE is 0.00 From the structural effect of O_DIV on RISK, the thesis found the direct impact of O_DIV on RISK with a marginal impact of -2.5%, and it is statistically significant O_DIV also demonstrates an indirect impact on RISK, but the marginal impact is shallow (0.334%) 29 Finally, the thesis also evaluates the structural effect of D_DIV on ROE Related to direct impact, the marginal impact of D_DIV on ROE is 9.3% Regarding indirect impact, the effect of D_DIV on ROE is 0.2% Then, the total impact of UA_DIV on ROE equals 9.5% The dissertation also investigated the structural effect of D_DIV on RISK, in which the direct effect is positive and statistically significant (2.063%), and the indirect effect equals 0.0063% (meager impact) As a result, the total impact of D_DIV on RISK is 2.063% 4.4.2 Recommendations First, to obtain better returns in the short term and at the same risk of bankruptcy, firms should focus on RADIV Increasing RA_DIV also increases RB_DIV then increase ROE Reducing UADIV does not reduce the ROE in the short run but reduces RISK In the long run, the impact of the investment may change the structural impact to ROE and RISK, especially the impact of UA_DIV Delay effect may occur in the model Therefore, firms need to monitor the historical data of investment to decide whether to expand or narrow each type of investment Second, owner diversification does not help to increase ROE in the short run, but it helps partly to reduce the risk of bankruptcy Therefore, when companies consider increasing the level of debt financing, it is necessary to consider raising equity capital to minimize the impact on the risk of debt financing 4.4.3 Limitations and further research 30 CHAPTER CONCLUSIONS AND RECOMMENDATIONS 5.1 Conclusions To investigate the impact of diversification on performance and risk, the thesis has conducted three research topics: Topic one: The impact of business diversification on performance and risk Topic two: The impact of asset investment diversification on business diversification, performance, and risk Topic three: The impact of capital diversification on asset investment diversification, performance, and risk From a theoretical point of view, the topic has some new contributions described as follows: Topic one: The impact of business diversification on performance and risk This topic is performed in chapter two Basically, the research in topic one is still a repetitive study of the impact of BDIV on performance and risk However, by measuring the risk of bankruptcy using the Merton-KMV model (probability of default model), and estimated these relationships using the SEM model, the research in the topic has contributed in terms of: (i) measuring the risk of bankruptcy by Merton-KMV model and (ii) employing SEM model to estimate empirically Topic two: The impact of ADIV on BDIV, performance and risk The research is addressed in chapter three The second topic considered ADIV as related and unrelated assets diversification and built the relationship between these two variables with BDIV, performance, and risk Thus, the second research topic has contributed regarding: (i) create new concepts (related and unrelated asset investment diversification) and (ii) build new link among concepts (ADIV, BDIV, performance, and risk) Topic three: The impact of CDIV on ADIV, performance, and risk The topic is organized in chapter four The research on the topic focused on understanding the CDIV The study analyzes CDIV as two components: equity diversification and debt diversification Then the thesis establishes the linkage of these two variables to the ADIV, performance, and risk Thus, similar to topic two, topic three has contributed to the following new points: (i) build new concepts (equity diversification and debt diversification) and (ii) establish new relationships among concepts (CDIV, ADIV, performance, and risk) The thesis then performs empirical estimation in each research topic, and the results are summarized as follows: Topic one: Research has used the GSEM estimation to replace the SEM estimation and find that the main research variable such as RB_DIV has a positive impact on ROE and RISK In contrast, the variable UB_DIV has a negative impact on both ROE and RISK Particularly regarding the impact on ROE, the marginal impact of the UB_DIV is higher than that of the RB_DIV and vice versa Finally, regarding the impact on RISK, the marginal impact of RB_DIV is higher than that of the UB_DIV 31 Topic two: Related to structural impact level of RA_DIV on ROE, RA_DIV has a direct and indirect impact on ROE; however, the total impact is shallow (0.03357) From the structural impact angle of RA_DIV on RISK, the thesis finds direct and indirect effects of RA_DIV on RISK, but the impact is also shallow (0.01063) The research also investigates the structural effect of UA_DIV on RISK, in which the direct effect is positive and statistically significant (equal to 0.062), but the indirect effect level is zero As a result, the total impact of UA_DIV on RISK is equal to the level of direct impact Topic three: The thesis also employs the GSEM estimate as a substitute for SEM In terms of the structural impact of O_DIV on ROE, the thesis explores the direct impact of O_DIV on ROE and the indirect impact of O_DIV on ROE through UA_DIV O_DIV does not show any direct impact on ROE Also, regarding indirect effects, the magnitude of the impact from UA_DIV on ROE is 0.00, so the overall impact level of O_DIV on ROE equals 0.00 Related to the structural impact of O_DIV to RISK, the thesis found direct and indirect effects of O_DIV on RISK, but the impact was shallow (-2.20%) For the structural effects of D_DIV on RISK, there are also direct and indirect effects 5.2 Recommendation Firstly, if firms expand their business to related business, the ROE will be improved but RISK will increase Conversely, expanding business to unrelated business will reduce RISK but also reduce ROE If the core business of the business is high risk, the business can consider the business diversification in the unrelated sectors to minimize RISK and accept the return as ROE will reduce If the business sector tends to go down or have a low level of risk, then the business should expand business to related industries On the whole, the impact of the two types of diversification to ROE and RISK is significant, but the tradeoff is not symmetrical Thus, firms in the normal business condition should be expanded to related business Secondly, to obtain better returns in the short term and at the same risk of bankruptcy, firms should focus on RADIV Increasing RA_DIV also increases RB_DIV then increase ROE Reducing UADIV does not reduce the ROE in the short run but reduces RISK In the long run, the impact of the investment may change the structural impact on ROE and RISK, especially the impact of UA_DIV Delay effect may occur in the model Therefore, firms need to monitor the historical data of investment to decide whether to expand or narrow each type of investment Finally, owner diversification does not help to increase ROE in the short run, but it helps partly to reduce the risk of bankruptcy Therefore, when companies consider increasing the level of debt financing, it is necessary to consider raising equity capital to minimize the impact on the risk of debt financing 5.3 Limitations The thesis has some limitations as follows: Firstly, the thesis has not studied the relationship among diversification, performance and risks in each sector Research in each sector can explore more detailed results and be more practical in practice; 32 Secondly, the thesis has not investigated the lagging effects It may appear the lagging impact of CDIV on ADIV, the lagging effect of ADIV on BDIV, or the lagging effect of BDIV on performance and risk Thirdly, the two output variables of the study (performance and risk) are only represented in turn by return on equity and bankruptcy risk Using only one variable representing each of the output variables may not reflect all views of performance and risk Fourthly, the topic introduces the new concept of RADIV and UADIV and simultaneously builds the structural theoretical relationship between among ADIV, performance and risk mediating variable of BDIV Although empirical research in the Vietnamese stock market has indicated the structural impact of RA_DIV on ROE and RISK via RB_DIV; however, there is no structural effect of UA_DIV to ROE and RISK because the UB_DIV variable does not become a mediating variable The empirical research effects need long-term research and larger sample size, and it needs to be studied in different countries to make firm conclusions in practice Finally, the thesis also introduces the new concept of owner diversification and debt diversification as well as builds the structural theoretical relationship between these variables to performance and risk through a mediator Although empirical research in the Vietnamese stock market has indicated the structural impact of D_DIV on ROE and RISK and the structural effect of O_DIV on RISK through the mediating variable RA_DIV, there is no structural effect of O_DIV to ROE These practical effects need longer-term research and larger sample size and need to be studied in different countries to make firm conclusions in practice 5.4 Further research Further studies may continue in the following directions: Firstly, follow-up studies may continue to examine the relationship between CDIV, ADIV, BDIV, performance and risk in different sectors Second, further studies may also extend variables that represent performance and risk for a more general perspective Third, it is necessary to extend empirical research in many different countries to the two types of variable proposed that are CDIV and ADIV Final, exploring the lagging relationship between the types of diversification and lagging impact of these types on performance and risk may be necessary ... CDIV on the performance and risk of firms through the mediating effect of ADIV 1.4 Object and scope of the study The object of the study is the impact of the diversification on the performance and. .. and risk of a firm More specifically, the impact of BDIV on the performance and risk of firms; the impact of ADIV on the performance and risk of firms through mediating effect of BDIV; and impact. .. BDIV on the performance and risk of firms Objective 2: Assess the impact of ADIV on the performance and risk of firms through the mediating effect of BDIV Objective 3: Evaluate the impact of the