The effect of cash flow sensitivity on enterprises cash holdings evidence from vietnam

Thus, finding the impact of cash flow sensitivity can help firms understand how cash flow affects their cash holdings, so they can provide better cash management and better implementatio

UNIVERSITY OF ECONOMICS ERASMUS UNVERSITY ROTTERDAM

HO CHI MINH CITY INSTITUTE OF SOCIAL STUDIES

VIETNAM THE NETHERLANDS

VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

THE EFFECT OF CASH FLOW SENSITIVITY

ON ENTERPRISES’ CASH HOLDINGS:

EVIDENCE FROM VIETNAM

BY

LE KHA TU

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, December 2017

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

HO CHI MINH CITY THE HAGUE

VIETNAM THE NETHERLANDS

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

THE EFFECT OF CASH FLOW SENSITIVITY

ON ENTERPRISES’ CASH HOLDINGS:

EVIDENCE FROM VIETNAM

A thesis submitted in partial fulfilment of the requirements for the degree of

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

By

LE KHA TU

Academic Supervisor:

Dr NGO MINH HAI

HO CHI MINH CITY, December 2017

ABSTRACT

Cash holdings have long been a particular concern when studying the financial policies of enterprises The global financial crisis in 2008 has impacted many economies in the world; Vietnam was one of the few emerging economies in Southeast Asia being less affected by the crisis However, the crisis has negative impacts on the macroeconomy of Vietnam Low GDP growth rates, booming inflation and escalating interest rates made Vietnamese enterprises’ situation more difficult Along with the Central bank's tightening monetary policy to curb inflation, commercial banks simultaneously restricted the issuance of large loans, which were more difficult for firms to access more capital In that context, the efficient use of cash holdings is the key to business success To use effectively the cash flow of the firms, it

is important to study the factors that affect the cash holdings of firms Thus, the author chose the topic "The effect of cash flow sensitivity on enterprises’ cash holdings: Evidence from Vietnam" to study the effect of financial factors on firms’ cash holdings in Vietnam The main objective of the study was to find empirical evidence

on the effects of cash flow sensitivities on cash holdings in the world and to seek evidence of this in Vietnam The study uses data from financial statements of 274 non-financial firms listed on Hanoi Stock Exchange (HNX) and Hochiminh Stock Exchange (HOSE) between 2009 and 2015 In Vietnam, cash flow sensitivities have a real impact on the cash holdings of firms, which is greater for firms with negative cash flows The financial constraint also affects the relationship between cash flow sensitivity and cash holdings, but the agency problem does not show any significant influence on this relationship Based on the results of the study, the paper offers some recommendations to help businesses make better cash management options At the same time, the paper also outlines some of the remaining limitations of research and future research

Keywords: Cash flow sensitivity, cash holdings, financial constraint, agency problem,

asymmetric

ACKNOWLEDGEMENT

The first word I would like to send my sincere thanks to all Vietnam – Netherlands Programme professors, who have taught and imparted me valuable knowledge during the time of study and research, enabling me to complete this thesis You all not only gave me the knowledge but also created the best conditions for me in the data collection and thesis writing process Friendly and kind Vietnam – Netherlands Programme staff were always willing to help me on physical facilities as well as references

I would like to sincerely and gratefully thank Dr Ngo Minh Hai, my supervisor, for his great assistance, principal and valued advice, guiding me from the smallest steps to help my thesis complete

I would like to express my deep gratitude to all my friends, families, who have always been with me, cheered and supported me all the time

Despite all the effort, due to limited knowledge and time, the problems presented in this thesis will certainly have many shortcomings I look forward to receiving

feedback, evaluations and comments from the professors

CONTENTS

CHAPTER 1: INTRODUCTION 1

1.1 The problem statement 1

1.2 The research objectives 2

1.3 Research contribution 2

1.4 The thesis structure 3

CHAPTER 2: LITERATURE REVIEW 4

2.1 Introduction 4

2.2 Studies on the effects of cash flow sensitivity on cash holdings 4

2.2.1 The impact of change in cash flow on cash holdings 4

2.2.2 The relationship between financial constraints and cash holdings of a company 7

2.3 Summary of previous findings 9

CHAPTER 3: RESEARCH METHODOLOGY 13

3.1 The analytical framework 13

3.2 The econometric models 14

3.2.1 Model the effect of cash flow sensitivity on cash holdings 14

3.2.2 The effect of cash flow sensitivity on cash holdings is under financial constraints 15

3.2.3 The effect of cash flow sensitivity on cash holdings is in the agency problem 17

3.3 The data 20

3.4 Fixed Effects and Random Effects 20

3.5 GMM4 estimations 21

3.6 Expectations on research results 22

3.6.1 Effects of cash flow sensitivity on cash holdings 22

3.6.2 Effects of cash flow sensitivity on cash holdings under financial constraints 23

3.6.3 Effects of cash flow sensitivity on cash holdings under agency problem 23

CHAPTER 4: RESEARCH RESULTS 24

4.1 Data description 24

4.2 Regression results and discussions 28

4.2.1 Effects of cash flow sensitivity on cash holdings 28

4.2.2 Effects of cash flow sensitivity on cash holdings under financial constraints 37

4.2.3 Effects of cash flow sensitivity on cash holdings under agency problem 42

CHAPTER 5: CONCLUSIONS 45

5.1 Conclusions 45

5.2 Policy implications 46

5.3 Limitations 47

REFERENCE 49

APPENDIX 51

LIST OF TABLES

Table 4 1: Sample statistics 24

Table 4 2: Variable description 25

Table 4 3: Compare the mean of variables 25

Table 4 4: Pearson and Spearman correlation coefficients 27

Table 4 5: The results of Almeida (2004) 29

Table 4 6: Estimation results of the model 31

Table 4 7: Results of the asymmetry sensitivity of cash flow 34

Table 4 8: Cash Flow and Cash of FPT Corporation for the period 2009 – 2015 36

Table 4 9: The number of observations of companies by year divided into three categories 38

Table 4 10: The summary statistics for the cash holdings of the two groups of companies by the three classification criteria 39

Table 4 11: Estimation results for financial constraints 40

Table 4 12: Estimation results of the model for agency problem 43

CHAPTER 1: INTRODUCTION

1.1 The problem statements

The 2008 financial crisis was the worst economic disaster since the Great Depression

of 1929, having a strong impact on the world in general and Southeast Asia in specific However, Vietnam was one of the few economies that were less affected by the recession caused by this crisis Nevertheless, the GDP growth rates for seven consecutive years (from 2009 to 2015) were in low level, showing that the crisis still had negative impacts on Vietnamese economy Moreover, in the period of 2009-2015, Vietnam’s inflation was abnormal (a sudden increase to 19% in 2011 and then declining), bad debts became a serious problem and getting worse Vietnamese enterprises were directly affected by the crisis, the number of enterprises reporting losses was increasing and many businesses were forced to dissolve In that context, cash holdings were paid more attention, since cash management is the key to a success business as accessing capital markets is difficult For effective cash management, the first problem is what factors affect the firm's cash holdings

Prior literature has widely studied the effects of cash flow on cash holdings and got certain conclusions Almeida et al (2004), Khurana et al (2006) found cash flow sensitivity was positive Meanwhile, Riddick and Whited (2009), Bao, Chan and Zhang (2012) found that cash flow sensitivity was negative

The study of cash flow sensitivity of cash holdings may help firms proposing a better cash management model, thereby firms use cash more efficiently However, in Vietnam, the effect of cash flow sensitivity on cash holdings has yet to be studied extensively, and there are disputations on this problem in the world For this reason, the author chose the research topic "The Effect of Cash Flow Sensitivity on Cash Holdings: Evidence from Vietnam", examining whether or not the effect of cash flow sensitivity on cash holdings in Vietnamese enterprises

1.2 The research objectives

The main purpose is examing the effect of cash flow sensitivity on cash holdings, the paper includes two basic objectives:

- Examining the effect of cash flow sensitivity on cash holdings in Vietnamese enterprises

- Verifying the nonlinear relationship between cash flow sensitivity and cash holdings

To accomplish two research objectives, the paper will solve the following research problems:

- Is there any evidence in the world about the effect of cash flow sensitivity on cash holdings?

- Which method is appropriate to test the effect of cash flow sensitivity on cash holdings in Vietnam?

- What factors affects the cash flow sensitivity on cash holdings?

- What is the effect of cash flow sensitivity on cash holdings in case the firm has positive cash flow and negative cash flow?

1.3 Research contribution

Research on the effect of cash flow sensitivities has a very important meaning, especially for firms Financially, cash flow sensitivity can represent a company's risk, because cash flow sensitivity is the level of change in the company's cash when cash flow changes If a company has high cash flow sensitivity, this can affect corporate financial policies such as dividends and capital structure A company with high cash-flow sensitivity is difficult to maintain a capital structure with high debt ratios due to insecure liquidity and the company is hard to implement a good dividend policy when cash always has instability Thus, finding the impact of cash flow sensitivity can help firms understand how cash flow affects their cash holdings, so they can provide better cash management and better implementation of dividend payment and capital mobilization policies The results of the study will also contribute to the global debate

on how cash flow sensitivity affects cash holdings

1.4 The thesis structure

The research is organized in the following chapters:

Chapter 1: Introduction This chapter presents the reasons for choosing topics, research objectives, research methods, research contribution and thesis structure

Chapter 2: Overview of previous studies on the impact of cash flow sensitivity on cash holdings This chapter presents the results of the previous study on the effect of cash flow sensitivity on cash holdings, findings, arguments and limited issues in these studies

Chapter 3: Research Methods This chapter will detail the research model, variables in the model, data as well as expectations about the research results

Chapter 4: Research Results This chapter presents and discusses the results of the study on the effect of cash flow sensitivity on cash holdings of firms in Vietnam, the results of examination with financial constraints and agency problem

Chapter 5: Conclusions and limitations of the study This chapter presents the contributions of the research, the next research direction, and the limitations

CHAPTER 2: LITERATURE REVIEW

to avoid the burden of costs One of the important cash-flowing channels for the company is through cash flow from its revenues The change in this cash flow will cause the change in the company's cash holdings Some researchers have focused on this aspect when studying cash holdings

2.2 Studies on the effects of cash flow sensitivity on cash holdings

2.2.1 The impact of change in cash flow on cash holdings

Almeida et al introduced a model of a firm’s liquidity demand that formalizes Keynes’ intuition The model hypothesizes that firms with capital constraints tend to have cash holdings from cash inflows Almeida et al argue that firms that were not constrained in their access to funding would not show cash incentives, when the firm’s cash flow changed, their cash holdings would not be affected while financially constrained firms would be affected by changes in cash flow

To test their hypotheses, Almeida et al used a database of manufacturing firms between 1971 and 2000, and used OLS regression to estimate the model The hypothesis of the model predicted that a change in cash holdings would correspond to the cash flow shock The hypothesis also predicted that the cash holdings of financially constrained firms would be affected by the attraction of future investment opportunities Therefore, Almeida et al use the Q variable to capture unobserved information about the value of long-term growth options

The OLS regression showed that for the limited-capital-access firms, when the cash flow was positive, the firm would increase cash holdings and vice versa Variable Q had a positive and significant coefficient for financially constrained firms With future

investment opportunities, the limited capital access firms would increase their cash holdings

Riddick and Whited (2009) validated cash flow sensitivity with different theories and models from Almeida et al Riddick and Whited argued that in the Almeida model the cash flow was positive because Almeida thought that an increase in cash flow did not correspond to higher yield of fixed assets So, when a company has a positive cash flow, the company will not have incentive to transfer its highly liquid assets to fixed assets, so the company will use increased cash flow to supplement its cash holdings

The results of Riddick and Whited differed from the results of Almeida et al It was caused by correction of the measurement error in the Tobin's Q variable Greene (1997, p 440) argued that the measurement error in Tobin's Q variable could affect cash flow variable because of the measurement error In a regression variable it affects all coefficients of variables in regression if this regression variable correlates with other variables

Riddick and Whited used the fourth-order GMM estimation to overcome the problem

of measurement errors in Tobin's Q The results showed that when a company has positive cash flow, its cash reserves would fall This happens because when there was

a positive yield shock that increased cash flow and margin profit of capital, the substitution effect would cause the company to use cash reserves to purchase more productive tangible assets and use cash to invest, so their cash reserves will decrease

Bao, Chan and Zhang (2012) also used the GMM4 estimation with additional empirical model to confirm the results of Riddick and Whited Bao, Chan and Zhang added to the experimental model of Riddick and Whited several limited control variables such as firm size, capital expenditure, non-cash flow, and short-term debt The results confirmed the results of Riddick and Whited that the relationship between cash flow sensitivity and cash holdings is negatively correlated

At the same time as testing the impact of cash flow sensitivity on cash holdings, researchers also concern that the relationship between cash flow and cash flow sensitivity is linear or nonlinear The linear relationship between a change in cash flow

the magnitude of change in cash reserves is unchanged According to Faulkender and Wang (2006), lacking cash would diversify the company's cash reserves This suggests that there are grounds to believe that the impact of cash flow sensitivity on cash flow will be different in the two cases where the firm has positive cash flow and the company has negative cash flow

Research results from Almeida et al (2004) indicated that firms were limited in their abilities to access funding, i.e the firms that were financially constrained, had positive cash flow sensitivity, but did not show the difference in the effect of cash-flow sensitivity on cash flow in two cases: the firm has positive cash flow and the firm has negative cash flow

Riddick and Whited (2009, p 1793) also pointed out briefly that medium and large firms exhibit negative nonlinear relationships in cash flow sensitivity, however, the study of Riddick and Whited did not focus on the nonlinear relationship between cash flow and cash flow sensitivity Riddick and Whited pointed out firms with an increase

in cash flow tended to hold cash on investments because positive cash flow was the evidence that tangible assets were more productive In addition, as cash flows increased, firms used cash reserves to invest in highly profitable projects Conversely, when a company had negative cash flow, this indicated that the tangible fixed assets were of low productivity, or the company had projects with negative NPV Then the company would immediately stop these projects to save cash Hence, according to Riddick and Whited, whether a company has a positive or negative cash flow, the relationship between cash flow sensitivity and cash holdings remains negative

Bao, Chan and Zhang (2012) argued that a company could not immediately stop the negative NPVs projects when the company has negative cash flow for three reasons First, some projects were accompanied by binding contracts and the company could not stop these projects immediately, this was common when these were tender projects The second reason was indicated by Kothari et al (2009) that managers had

an incentive to hide bad information If a company immediately terminates a bad project, then maybe the market will realize the problem in that company So, some managers may choose to keep bad projects to minimize bad news By delaying the

release of bad news, managers expect to be able to extend the time until there is good information to neutralize the impact of bad news Therefore, when a company faces negative cash flow, the company may not end up ineffective projects The third was the agency problem, Jensen and Meckling (1976), Jensen (1986) pointed out that managers had an incentive to invest excessively and therefore they could keep some projects with negative NPV to maximize personal profit This is similar to the risk transfer in case of financial exhaustion, so when the company has negative cash flow, the cash reserves of the company may not increase but may decrease as managers continue to bring cash to invest in projects with negative NPV

Because of three above reasons, a company facing negative cash flow may not immediately stop all bad projects, when the company has negative cash flow, the company's cash reserves may not increase Using the samples of manufacturing firms from 1972 to 2006, Bao, Chan and Zhang examined animpact of cash flow sensitivity

on cash flows including the viability of the nonlinear relationship corresponding to a change in cash holdings between the two cases where the firm had positive cash flow and the firm has negative cash flow Empirical evidence suggested that cash flow sensitivity would be negative when the firm had positive cash flow, in line with Riddick and Whited (2009), but the sensitivity of cash flow would be positive when the firm had a positive cash flow, this supports the hypothesis of Bao, Chan and Zhang that the influence of cash flow sensitivity on cash holdings is asymmetrical

2.2.2 Relationship between financial constraints and cash holdings of a company

Two areas of important research in corporate finance are the impact of financial constraints on corporate policy and how firms manage finance These two areas have a close relationship If a company has unlimited access external funding, which means the firm is not financially constrained, the company does not need to have cash reserves for future investment On the contrary, if a company is limited in accessing external funding because of high cost, cash reserves are necessary to cover the needs

of the company

According to Kaplan and Zingales (1997), the most accurate and most widely used definition of financial constraints firms was that these firms distinguish the internal

and external capital expenditure However, by definition, every company seems to be financially constrained A small transaction cost of using external funding is sufficient

to rank the company into financial constraints Fazzari, Hubbard and Petersen (1988) and many other researchers considered that the cash-flow sensitivity of the higher investment was the evidence for larger financial constraints

In their study, Almeida et al (2004) demonstrated that the relationship between financial constraints and the company's payment needs could help determine whether financial constraints were an important determinant of corporate behavior or not Almeida argued that holding cash was expensive, as higher cash savings would require cuts from profitable investment projects Therefore, financially constrained firms choose an optimal cash policy to balance the returns of present and future investment projects This policy is the opposite of firms that are not financially constrained: firms which are not financially constrained neither use cash nor face the cost of cash holdings

Almeida uses five methods to classify firms into two group of financially constrained and un-constrained firms: dividend payment policy, asset size, bond ratio, percentage

of commercial papers and KZ (based on Kaplan and Zingales (1997))

Almeida et al (2004) concluded that changes in cash flow affected cash holdings only when firms were constrained in access to capital, for firms with no financial constraints, changing cash flows would not affect cash holdings This may be caused

by financially constrained firms tending to hold cash, because of the limited capital access, while financially un-constrained firms do not

Riddick and Whited (2009) predicted that firms with higher external capital expenditure would have a greater negative correlation coefficient The results of the regression of Riddick and Whited did not support the opinion of Almeida et al in which financial constraint firms did not have cash flow sensitivities of holding cash, and even did not support Riddick and Whited’s prediction stated above

The research results of Bao, Chan and Zhang (2012) supported Riddick and Whited (2009) that cash-flow sensitivity was stronger for medium- and large-sized firms because these firms had no financial constraints

2.3 Summary of previous findings

Although previous studies use different empirical models and methods, most of the studies conclude that cash flow sensitivity has an effect on cash holdings The discussed problem is how this effect tends to be Almeida et al (2004) used the OLS method for data taken from US manufacturing firms in the period 1971-2000 The positive relationship between changes in cash holding and cash flow indicates that firms decrease (or increase) cash reserves when they have negative (or positive) cash flows A change of cash flow only affects cash holdings for firms with financial constraints Almeida et al found that their findings were consistent with their argument and argued that only financially constrained firms, because of their restricted access to external capital markets, saved cash out of cash flow, while unconstrained firms did not In the same direction of research with Almeida, Khurana

et al (2006) also drew the same conclusion

In normal intuition, the conclusion of Almeida et al seems to be right, when a firm has a positive cash flow i.e profitable business operations, the firm will bring that profit back into its cash reserves However, some researchers disagree with Almeida's argument Riddick and Whited (2009) argued that the results of Almeida showed positive cash flow sensitivity because in Almeida's model, the increase in cash flow did not correspond to the increase in productivity of physical assets Riddick and Whited contended that a firm with an increase in cash flow tended to turn cash reserves into investments because the positive cash flow shock indicated higher productivity of physical assets Riddick and Whited thought the positive cash flow sensitivity was not appropriate because of the OLS method Riddick and Whited pointed out that due to the measuring error of Tobin's Q variable, using the OLS method would affect the coefficient of other variables in the estimation, especially the cash flow variable Hence, Riddick and Whited used a general method of moment (GMM) estimation, proposed by Erickson and Whited (2000), to estimate the model The results were consistent with Riddick and Whited's original hypothesis, the cash flow sensitivity was negative, which meant that firms with an increase in cash flow tended to turn cash reserves into investments, reducing firm's cash holdings On the

Almeida et al Almeida said that firms were not financially constrained, their cash was not affected by changes in cash flow, Riddick and Whited proved that firms without financial constraints had cash flow sensitivity, and even the cash flow sensitivity of these firms was greater than that of financially constrained firms

Sharing the same empirical method with Riddick and Whited (2009), suggesting that the fourth-order GMM (GMM4) was more suitable than OLS, Bao, Chan and Zhang (2012) performed a reassessment of the effect of cash flow sensitivity on cash holdings and the research results support the outcome of Riddick and Whited However, Bao, Chan, and Zhang went further than Riddick and Whited to pay attention to a nonlinear relationship between cash flow sensitivity and cash holdings Bao, Chan and Zhang argued that cash flow sensitivity was not negative in all cases, particularly when the firm had negative cash flow Riddick and Whited argued that when a firm had negative cash flow it meant the productivity of the firm's physical assets was decreasing or the firm was investing in ineffective projects Then the firm would stop funding projects or stop procuring assets, which would increase the firm's cash Bao, Chan and Zhang in their study gave some main reasons why the firm could not immediately stop ineffective projects, therefore, when firm had negative cash flow, the negative cash flow sensitivity may not be sustainable The empirical results confirmed the predictions of Bao, Chan and Zhang

Besides the studies on the effects of cash flow sensitivity on cash holdings, studies on impacts of agent costs on the effect of cash flow sensitivity on cash holdings had certain results Dittmar, Mahrt-Smith, and Servaes (2003) argued that if firms have a larger agency problem, firms would hold more cash Bao, Chan and Zhang (2012) found a link between external control and the effect of cash flow sensitivity on cash holdings If the firm's shares were held by institutional investors, i.e the firm had a large external control and a lower agency, the effect of cash flow sensitivity on cash holdings would be smaller

CHAPTER 3: RESEARCH METHODOLOGY

3.1 The analytical framework

Firm performance

Increasing investment

Increasing tangible

assets

Agency problem

Cost of holding money

Investment opportunities

The difficulties in accessing capital

External capital cost

The sensitivity

of Cash flow

Financial constraint

Cash holdings

3.2 The econometric models

Based on research objectives, the paper has three problems to clarify The first is the influence of cash flow sensitivity on cash holdings, the second is the impact of financial constraints the last is the agency cost Therefore, the research paper uses three models to solve the above problems

3.2.1 Model the effect of cash flow sensitivity on cash holdings

Based on the model of Bao, Chan and Zhang (2012), the paper uses the following model to test the hypothesis that cash flow sensitivity affects cash holdings and this effect is different in case the firm has negative and positive cash flow:

ΔCashHoldings it = α 0 + α 1 CashFlow it + α 2 Neg it + α 3 Cashflow it * Neg it + α 4 Q it +

α 5 Size it + α 6 Expenditure it + α 7 Acquisition it + α 8 ΔNCWC it + α 9 ShortDebt it-1 + ε it (1) Where:

CashHoldings: is calculated as the cash in company divided by total asset

ΔCashHoldings: the difference of cash between year t and year t−1 over total assets CashFlow: is the profit after interest, dividends and taxes payments plus depreciation

over total assets

Neg: is the indicator variable Neg equals zero if in that year the firm has positive cash

flow and one otherwise

Q: represents the market capitalization of the company Q is calculated by total of the

market value of the capital and the book value of assets minus the book value of the capital and divided by the book value of assets

Size: represents the scope of the company, calculated as the natural logarithm of total

assets

Expenditure: the ratio of capital expenditures to total assets

Acquisition: the indicator variable If the company has no acquisition in that year, Acquisition equals one and zero otherwise

NCWC: is calculated as net non-cash working capital which equals working capital

minus cash divided by total assets

ΔNCWC: is the difference of NCWC between year t and year t−1

ShortDebt: is the debt in short-term weighted by total assets

i and t refers to firm and year respectively, ε is random error term

3.2.2 The effect of cash flow sensitivity on cash holdings is under financial constraints

After examining the effect of cash flow sensitivity on cash holdings, the study examines how cash flow sensitivity varies between financially constrained and un-constrained firms For that purpose, the paper uses three measures to divide the sample into two groups

Firstly, the paper uses the KZ index, which is based on the results of Kaplan and Zingales (1997) The study uses the KZ index as the data for this metric corresponds

to the data in Vietnam The KZ index is calculated as follows:

KZindex = -1.002 x CashFlow + 0.283 x Q + 3.139 x Leverage - 39.368 x Dividends - 1.315 x CashHoldings

Where:

CashFlow: is the profit after interest, dividends and taxes payments plus depreciation

over total assets

Q: represents the market capitalization of the company Q is calculated by total of the

market value of the capital and the book value of assets minus the book value of the capital and divided by the book value of assets

Leverage: represents the capital structure of the firm which is calculated as the ratio of

the total debt and total assets

Dividends: represents the dividend policy of the company which is calculated as the

cash dividends divided by total assets

CashHoldings: is calculated as the cash in company divided by total asset

The study evaluates firms via the KZ index based on Almeida et al (2004), within a year, firms ranked among the 33% of the highest KZ index are considered as in the financial constraint group

Secondly, the study uses an additional index to assess the financial constraints among firms This index is based on the research by Whited and Wu (2006) and is called the

WW index Bao, Chan, and Zhang (2012) argued that the WW was more appropriate than the KZ as firm’s characteristics in the WW were more related to the firm's financial constraints than the KZ The WW is more relevant than the KZ since the

WW does not include Tobin's Q The data for calculating the WW are also consistent with the data in Vietnam The WW index is calculated as follows:

WWindex = -0.091 x CashFlow it - 0.062 x DIVPOS it + 0.021 x TLTD it – 0.044 x Size it

+ 0.102ISG it – 0.035 x SG it

Where:

CashFlow: is the profit after interest, dividends and taxes payments plus depreciation

over total assets

DIVPOS: the indicator variable If firm i pays dividend by cash in year t, DIVPOS is

consider as one and zero otherwise

TLTD: is the debt in long-term weighted by total assets

Size: represents the scope of the company, calculated as the natural logarithm of total

assets

ISG: the industry's revenue growth rate

SG: the growth rate of the company

According to Bao, Chan and Zhang (2012), in a year, firms in the top 25% of the highest WW index were considered as firms in the financial constraint group

Finally, the study assesses whether firm is assigned to the financially constrained group or not based on dividend payments If a firm does not pay dividends during the year, then the firm is considered financially constrained firms and vice versa

After dividing the sample into two groups, the equation (1) is adjusted to examine the effect of financial constraints on the firm's cash holdings The study is modified equation (1) into equation (2) as follows:

ΔCashHoldings it = β 0 + β 1 CashFlow it + β 2 Neg it + β 3 Cashflow it * Neg it +

β 4 Constraintit + β 5 CashFlow it * Constraint it + β 6 Constraint it * Neg it + β 7 CashFlow it * Constraint it * Neg it + β 8 Q it + β 9 Size it + β 10 Expenditure it + β 11 Acquisition it +

β 12 ΔNCWC it + β 13 ShortDebt it-1 + ε it (2) The variables in equation (2) are defined as in equation (1)

In equation (2), there is a constraint dummy variable (value of one if the firm is considered financially constrained) and Constraint's interactive variables with CashFlow variable, Neg dummy variable

3.2.3 The effect of cash flow sensitivity on cash holdings is in the agency problem

To consider the impact of the agency problem, the study uses the following model:

ΔCashHoldings it = γ 0 + γ 1 CashFlow it + γ 2 Neg it + γ 3 Cashflow it * Neg it + γ 4 Inst it +

γ 5 CashFlow it * Inst it + γ 6 Inst it * Neg it + γ 7 CashFlow it * Inst it * Neg it + γ 8 Q it + γ 9 Size it

+ γ 10 Expenditure it + γ 11 Acquisition it + γ 12 ΔNCWC it + γ 13 ShortDebt it-1 + ε it (3)

Where: Inst equals one if the number of shares held by institutional shareholders is in

the top 10% of the company

The variables in the model are calculated as follows:

Cash holdings, in the study of Almeida et al (2004) on cash flow sensitivity, Almeida

et al defined the firm's cash holdings including cash and marketable securities divided

by total assets Bao, Chan and Zhang (2012) defined cash as the firm’s total cash holdings divided by total assets With data taken from Vietnameses firm’s financial statements, the study calculates cash holdings as cash and cash equivalents on the balance sheet, i.e the largest liquid asset item, divided by the total assets of the firm

𝐶𝑎𝑠ℎ𝐻𝑜𝑙𝑑𝑖𝑛𝑔 = 𝐶𝑎𝑠ℎ 𝑎𝑛𝑑 𝑐𝑎𝑠ℎ 𝑒𝑞𝑢𝑖𝑣𝑎𝑙𝑒𝑛𝑡𝑠

𝑇𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠

CashHoldings: is calculated as the cash in company divided by total asset

Cash flow, Almeida et al (2004), Bao, Chan and Zhang (2012) defined the firm’s cash

flow as earnings before extraordinary items and depreciation divided by total assets Extraordinary items or expenses are stated in the income statement However, with data collected in Vietnam, there are no extraordinary items in the firms' income statement Therefore, the study takes the variable definition according to Bates et al (2009), in which CashFlow variable is the profit after interest, dividends and taxes payments plus depreciation By this definition, the CashFlow variable is calculated as follows:

𝐶𝑎𝑠ℎ𝐹𝑙𝑜𝑤 = 𝑃𝑟𝑜𝑓𝑖𝑡 𝑎𝑓𝑡𝑒𝑟 𝑡𝑎𝑥 𝑝𝑙𝑢𝑠 𝑑𝑒𝑝𝑟𝑒𝑐𝑖𝑎𝑡𝑖𝑜𝑛

𝑇𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠Where: depreciation is taken from the depreciation of fixed assets on the indirect cash flow statement

Price to book value, known as the margin value q, was developed by Tobin (1969) In

this study, Tobin defined the margin value q as the ratio between the market value and the replacement value of the asset However, since the margin value q can not be observed, the study uses the value of Tobin's Q to replace the margin value Q, which

is used in many empirical studies in the world such as Bates et al (2009), Bao, Chan and Zhang (2012) The value of Tobin's Q is calculated by comparing the firm’s

market capilization value with the respective book value

𝑄 = 𝑀𝑎𝑟𝑘𝑒𝑡 𝑐𝑎𝑝𝑖𝑙𝑖𝑧𝑎𝑡𝑖𝑜𝑛 + 𝐵𝑜𝑜𝑘 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑑𝑒𝑏𝑡𝑠

𝑇𝑜𝑡𝑎𝑙 𝑏𝑜𝑜𝑘 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑎𝑠𝑠𝑒𝑡𝑠Book value of debts is taken from liabilities item on balance sheet

Market capilization equals P/B multiplied by book value of the capital ie equals outstanding shares multiplied by price at that time

Total book value of assets is taken from total assets item on balance sheet

Capital expenditure, Riddick and Whited (2009) argued that as firm's cash flow

increased, it corresponds to an increase in the productivity of firm's tangible fixed assets Firms do not accumulate cash, they return cash to invest It means firms use cash to purchase more fixed assets when they have positive cash flow The amount of money a firm spending on procuring fixed assets for investment projects is called capital expenditure, which suggests that capital expenditures may represent investment opportunities Based on this argument, the study calculates capital expenditures as the difference between the firm's fixed assets in year t and t -1 plus the

fixed asset depreciation in year t and t - 1

𝐸𝑥𝑝𝑒𝑛𝑑𝑖𝑡𝑢𝑟𝑒 = 𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑜𝑓 𝑓𝑖𝑥𝑒𝑑 𝑎𝑠𝑠𝑒𝑡𝑠 + 𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑜𝑓 𝑑𝑒𝑝𝑟𝑒𝑐𝑖𝑎𝑡𝑖𝑜𝑛

𝑇𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠

Non-cash working capital, based on the definition of Bao, Chan and Zhang (2012),

non-cash working capital is calculated as working capital minus cash divided by total assets Working capital equals short-term assets minus short-term debts Cash

holdings equal cash and cash equivalents

𝑁𝐶𝑊𝐶 = 𝑆ℎ𝑜𝑟𝑡 𝑡𝑒𝑟𝑚 𝑎𝑠𝑠𝑒𝑡𝑠 − 𝑆ℎ𝑜𝑟𝑡 𝑡𝑒𝑟𝑚 𝑑𝑒𝑏𝑡𝑠 − Cash and cash equivalents

𝑇𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠

ΔNCWC: is the difference of NCWC between year t and year t−1

Short-term liabilities, according to Bao, Chan and Zhang (2012), ShortDebt variable is

calculated by taking short-term debts in year t-1 divided by total assets in year t

𝑆ℎ𝑜𝑟𝑡𝐷𝑒𝑏𝑡 = 𝑆ℎ𝑜𝑟𝑡 𝑡𝑒𝑟𝑚 𝑑𝑒𝑏𝑡𝑠 𝑝𝑟𝑒𝑣𝑖𝑜𝑢𝑠 𝑦𝑒𝑎𝑟

𝑇𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠

Size, Bates et al (2009) argued that there was economic benefit to the size of cash

holdings In this study, Bates defined firm size as the natural logarithm of the firm's total assets This definition is also used in many other studies, such as Almeida et al (2004), Bao, Chan and Zhang (2012)

𝑆𝑖𝑧𝑒 = ln (𝑇𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠)

3.3 The data

Based on previous studies on cash holdings, the study selected data including 274 firms from the Hanoi Stock Exchange (HNX) and Hochiminh Stock Exchange (HOSE) Selected firms are non-financial The study excludes firms in the financial sector, such as the real estate, securities and banking, insurance due to special accounting standards, according to Shadi Farshadfar and Reza Monem (2013) Sampling period is from 2009 to 2015 The data is presented panel data with 1918 observations Data is collected from the financial statements of the firms which are taken from Vietstock, CafeF

3.4 Fixed Effects and Random Effects

The study uses the OLS regression method for the data panel as Fixed effects and Random effects The study ran two quantitative methods simultaneously with the collected data After running the regression, the study will use the Hausman test to examine the results from the two methods to find the better results

Hausman's test examines the results between the two regression models Fixed effects and Random effects by examining the difference in the regression coefficient and the null hypothesis assuming that the difference is not systematic That is, if accepting the hypothesis H0, the random effects are better In contrast, rejecting the null hypothesis, the Fixed effects are better

If the Hausman test‘results indicated that FEM is better REM, the study will use the results of the FE model However, to eliminate the heteroskedasticity, the study will use Robust and Cluster for the FE model

Robust, when there is the heteroskedasticity, OLS estimation is still an unbiased

estimator However, estimates are no longer effective because the variance is no longer the smallest Furthermore, the variance estimates will be biased, so that

significance level and confidence interval tests based on t and F are no longer reliable

Derived from the idea that the OLS estimation when there is the heteroskedasticity, the regression coefficient is correct, only the standard error is deviated, with the

Robust adjustment method, the regression coefficient will be retained, and the variance will be adjusted to achieve the smallest value

Cluster, to fix the autocorrelation, the study uses cluster-adjusted By fixing the id

value (representing for companies), this measure will fix the autocorrelation phenomenon in the variance

When estimating with FE adding Robust and Cluster, the results will be better, and the forecasting will be better

3.5 GMM4 estimations

One of the methods to solve the problem of error in variable estimation is to use instrumental variables in the model When the model has endogenous variables, or the variables in the model are correlated with the error of the model, the use of instrumental variables is necessary to eliminate the variance in the estimation of the model Instrumental variables must be correlated with endogenous variables in the model and not be correlated with model errors Therefore, in this paper, the author will use another method proposed by Erickson and Whited (2000), which is high-order GMM estimation

Erickson and Whited proposed a method for estimating models including accurately measuring variables and measurement errors, without the use of tool variables, which was the use of available information in high order moments of variables in the model

If F is a random variable, and k is a natural number, then E (Fk) is called kth order moment of F, and the quantity E (F - E (F)) k is called kth order center moment of F Each moment of the random variable represents a distributed characteristic of the random variable The first-order moment of F is the expected value of the variable; the second-order center moment of F is the variance

The idea of the high-order GMM estimation is to use the information contained in the high-order moment of the explanatory variables to calculate the regression coefficients, rather than directly using the value of the variable to calculate, to avoid deviation due to variable measurement error The effectiveness of the GMM4 estimation was demonstrated by the Monte Carlo simulation model of Erickson and

Whited (2000), as well as in some recent studies The high-order GMM estimation is considered to be able to overcome the problem of measurement errors of Tobin's Q

In this study, the author will use FEM, REM and fourth-order GMM as the estimation method for the research model Based on the results of the regression with data taken from Vietnam, the paper will select a suitable method to be the main estimation for the research model

3.6 Expectations on research results

3.6.1 Effects of cash flow sensitivity on cash holdings

In the test equation for the effect of cash flow sensitivity on cash holdings:

ΔCashHoldings it = α 0 + α 1 CashFlow it + α 2 Neg it + α 3 Cashflow it * Neg it + α 4 Q it +

α 5 Size it + α 6 Expenditure it + α 7 Acquisition it + α 8 ΔNCWC it + α 9 ShortDebt it-1 + ε it (1) Based on the results of previous studies, α 1 coefficient of the CashFlow variable is expected to be positive in the OLS regression, consistent with the results of Almeida

et al (2004), Almeida believed that firms would save cash when the company had a positive cash flow to prepare for future needs With the GMM4 regression, the results

of previous studies agreed that the α 1 coefficient would be negative, indicating that cash flow sensitivity is negative However, given the fact that Vietnam's data in the study is in the post-crisis period, within short sampling period, the study suggests that the negative impact of cash flow may not be the case for firms in Vietnam Thus,

expectation of the study on the α 1 coefficient can be negative or positive The α 3

coefficient is expected to be positive, as firms with negative cash flow will use cash

reserves to continue financing current projects A positive and significant α 3

coefficient in the case of a negative α 1 coefficient would indicate an imbalance in the impact of cash flow sensitivity on cash holdings

The coefficients of the two control variables Expenditure and Acquisition are expected

to bear a negative sign as capital spending and mergers will reduce the company's

cash holdings The coefficient of the ShortDebt variable can be either positive or

negative, as this variable shows the cash flow going out during the year, which can reduce the cash or increase the cash flow of the manager

3.6.2 Effects of cash flow sensitivity on cash holdings under financial constraints ΔCashHoldings it = β 0 + β 1 CashFlow it + β 2 Neg it + β 3 Cashflow it * Neg it +

β 4 Constraintit + β 5 CashFlow it * Constraint it + β 6 Constraint it * Neg it + β 7 CashFlow it * Constraint it * Neg it + β 8 Q it + β 9 Size it + β 10 Expenditure it + β 11 Acquisition it +

β 12 ΔNCWC it + β 13 ShortDebt it-1 + ε it (2)

The β 1 coefficient is expected to be negative or positive, the β 3 coefficient is expected

to be positive due to the effect of cash flow sensitivity on cash and asymmetry of cash

flow sensitivity is expected to remain under control of the financial constraints The β 5

coefficient is expected to be positive because financial constraints are assumed to be due to limited access to capital, which will ignore good investment opportunities in

order to increase cash savings when the firm has positive cash flow As a result, the β 7

coefficient is expected to be negative as financial constraint firms will find it difficult

to maintain existing projects with cash reserves rather than financial unconstraint firms

3.6.3 Effects of cash flow sensitivity on cash holdings under agency problem

ΔCashHoldings it = γ 0 + γ 1 CashFlow it + γ 2 Neg it + γ 3 Cashflow it * Neg it + γ 4 Inst it +

γ 5 CashFlow it * Inst it + γ 6 Inst it * Neg it + γ 7 CashFlow it * Inst it * Neg it + γ 8 Q it + γ 9 Size it

+ γ 10 Expenditure it + γ 11 Acquisition it + γ 12 ΔNCWC it + γ 13 ShortDebt it-1 + ε it (3)

Inst variable represents the volume of institution’s shares in firms The volume of

share holding of institutions in large firms represents low representation In this

equation, the γ 1 coefficient is expected to be negative or positive, while the γ 3

coefficient is expected to be positive because the effect of cash flow sensitivity on

cash holdings is expected to hold in the presence of the agency cost The γ 5 coefficient

is expected to be negative because firms have more outside control, which corresponds to lower agency costs; when there is positive cash flow, the company will

not accumulate cash but will use more cash to finance the projects While the γ 7

coefficient is expected to be negative, for firms with negative cash flows, more organizations holdings will prevent managers from over-investing in inefficient projects to seek personal profit, so the cash holdings of these firms tend to increase

CHAPTER 4: RESEARCH RESULTS

4.1 Data description

Table 4 1: Sample statistics

(This table shows the number of firms in each industry as well as the percentage of firms in each industry in the sample)

Table 4 2: Variable description

ΔCashHoldings 1,918 0.030 0.011 0.109 -0.014 0.065 Cashholdings 1,918 0.117 0.076 0.124 0.028 0.165

Expenditure 1,918 0.018 0.023 0.137 -0.022 0.048 Acquisition 1,918 0.051 0.000 0.219 0.000 0.000

Constraint Unconstraint Difference t statistics

the book value of assets, Size is the natural logarithm of the total asset, Expenditure is the ratio of capital expenditures to total assets, Acquisition is an indicator variable that equals one if the firm makes an acquisition in that year and zero otherwise, ΔNCWC is the difference in non-cash working capital between year t and year t−1, ShortDebt is the debt in short-term weighted by total assets)

Table 4.2 and 4.3 presents descriptive statistics of variables Table 4.2 shows the descriptive statistics of the variables in the model, while Table 4.3 compares the mean

of variables between the two groups of companies classified by the WW index Table 4.2 shows that the mean and median of the ΔCashHoldings variables are respectively 0.030 and 0.011, indicating that only small changes in the cash holdings of the companies in the sample Cash accounts for an average of 11.7% of total assets of companies Meanwhile, short-term debt accounted for an average of 38.5% of total assets, indicating that the debt ratio of Vietnamese companies was quite high The company's non-cash working capital changes a little with a mean of 0.162 and a median of 0.018 The rate of mergers and acquisitions is quite low, only about 5.1% of the companies in the sample have acquisitions or mergers

In Table 4.3, the criteria such as cash holdings, acquisition, working capital and term debt are not significant difference between Financial unconstraint firms also undertake more mergers and acquisitions, along with larger capital expenditures, which are understandable because the group is easier to access funds than financial constraint firms The Q coefficients of firms that are not financially constrained also show that they have more growth opportunities

short-Table 4 4: Pearson and Spearman correlation coefficients

(This table presents the correlation coefficients between the variables in the model Correlation coefficients include the Pearson coefficient (the part below the diagonal) and the Spearman coefficient (the part above the diagonal))

ΔCashHoldings CashFlow Q Size Expenditure Acquisition ΔNCWC ShortDebt

Table 4.4 presents Pearson and Spearman correlation coefficients between variables in the model The correlation coefficient between Q and CashFlow is 0.178, which is greater than other variables However, the correlation coefficient between Q and CashFlow is not high, suggesting that the Q error would not significantly effect on the coefficient of CashFlow on the OLS regression The correlation coefficient of ΔCashHoldings and CashFlow variables are positive and significant, indicating that increases in cash flow tend to increase the cash holdings of companies CashFlow and Expenditure have a positive correlation, indicating that when the company has increased cash flow, the company will invest in new projects, which will be further verified in the model estimation results

4.2 Regression results and discussions

4.2.1 Effects of cash flow sensitivity on cash holdings

To test the relationship between cash and cash flow sensitivity, the study will use the Almeida et al (2004) model to compare the results with the main model of the study Compared to the main model, Almeida et al model has fewer control variables The model of Almeida et al (2000) is as follows:

ΔCashHoldings it = α 0 + α 1 CashFlow it + α 2 Q it + α 3 Size it + ε it

Riddick and Whited (2009) shown that when an explanatory variable has a measurement error, the estimates of other explanatory variables in the model may be affected Toni and Whited (2000) developed a high-order GMM to overcome the problem of measurement error in estimation The study will apply FEM, REM and GMM4 to estimate the model

Table 4 5: The results of Almeida (2004)

Hausman specification test between FEM and REM χ2 (df =3) = 18.23

(Samples include non-financial companies from 2009 to 2015 The dependent variable ΔCashHoldings is the difference of cash between year t and year t−1 over total assets, CashFlow is the profit after interest, dividends and taxes payments plus depreciation over total assets, Q is calculated by total of the market value of the capital and the book value of assets minus the book value of the capital and divided by the book value

of assets, Size is the natural logarithm of the total asset ***, **, * are respectively significant at 1%, 5%, 10%)

In the REM method, the CashFlow coefficient is positive and significant, consistent with the results of Almeida et al (2004), suggesting that when the cash flow of the firm increases, the cash holdings also increased The Q coefficient is positive and significant, which is consistent with the results of Almeida et al., In Almeida's model, the Q variable represents long-term growth options, when there is growth opportunity, the cash holdings of the company will increase to prepare for new investment projects

Using the Hausman test to determine the resultant choice between FEM and REM models, the results were significant, and the null hypothesis was rejected, which meant that the FEM model had better results However, according to the FEM method, the CashFlow variable was positive and not statistically significant

The GMM4 method is used to estimate the model of Almeida et al Riddick and Whited found that the sensitivity of cash flow was negative when using the GMM4 estimation method to eliminate the measurement error in the Q variable However, in GMM4 estimation for the Almeida model with data in Vietnam, the sensitivity of the cash flow was positive and not significant, similar to the results of the FEM method

In order to test whether this is a characteristic of Vietnamese companies or because the model of Almeida is not suitable, the study continues to estimate the equation (1)

to test the effect of sensitivity cash flow on cash holdings, temporarily disregarding the asymmetric of the cash flow sensitivity by removing the dummy variable Neg and the interaction variable CashFlow * Neg GMM4, FEM and REM were used concurrently The equation (1) of the study does not take into account the existence of

a nonlinear relationship between cash flow sensitivity and cash holdings:

ΔCashHoldings it = α 0 + α 1 CashFlow it + α 2 Q it + α 3 Size it + α 4 Expenditure it +

α 5 Acquisition it + α 6 ΔNCWC it + α 7 ShortDebt it-1 + ε it

Table 4 6: Estimation results of the model

Hausman specification test between FEM and REM χ2 (df =7) = 22.36

(The sample includes non-financial companies from 2009 to 2015 The dependent variable ΔCashHoldings is the difference of cash between year t and year t−1 over total assets, CashFlow is the profit after interest, dividends and taxes payments plus depreciation over total assets, Q is calculated by total of the market value of the capital and the book value of assets minus the book value of the capital and divided by the book value of assets, Size is the natural logarithm of total assets, Expenditure is the ratio of capital expenditures to total assets, Acquisition is indicator variable 1 if the company has acquisitions during the year, ΔNCWC is the non-cash flow difference divided by total assets, ShortDebt is the debt in short-term weighted by total assets

***, **, * respectively significant at 1%, 5%, 10%)

In the regression results of GMM4, FEM and REM, cash flow sensitivity was positive, consistent with the finding of Almeida et al (2004) Capital expenditure is negative, which is a good thing because the company's cash will decrease when the company spends for purchasing fixed assets The coefficient of ShortDebt is negative, meaning that in Vietnam, when the debt ratio increases, firm gross profit will be deducted by a huge additional amount of interest because of high debt ratio As a result, the company's after-tax profit will be reduced, and cash will decrease accordingly On the other hand,

if a company has a low short-term debt ratio, profit will be used less to pay interest, so the cash holdings of the firm will also be higher Normally, when an enterprise wants to increase working capital, there are two ways in which an enterprise can increase its short-term debt to increase its cash inflows or sell its long-term assets for investing in short term assets With the first option, increasing short-term assets by borrowing, in the long run, will lead to short-term liabilities increase, then the increase of short-term assets by debt does not seem to change working capital, even it also reduces the liquidity of the company Therefore, it is more appropriate for businesses to sell their long-term assets for increasing their short-term assets Sales of long-term assets will increase the working capital of the business, and growth of non-cash working capital will be faster than increases in cash holdings because with this option, the business tends to invest in non-cash short-term assets such as inventory, tools, equipment rather than to increase cash in the company

However, the coefficient of CashFlow in the GMM4 estimator is not significant, which may be a sign that the GMM4 is no longer a suitable estimation method for data in Vietnam

Using the FEM and REM method, the results show that the Q coefficient is positive and statistically significant at 1%, the coefficient of CashFlow variable is positive and statistically significant at the significance level of 5 % and 10%, indicating that the FEM and REM methods are more appropriate than the GMM4 method for the data of Vietnamese enterprises Using the Hausman test to determine the resultant choice between the two FEM and REM, the results were statistically significant, and the null hypothesis was rejected, which meant that the FEM model had better results The

coefficient of CashFlow variable in FEM regression is positive, in line with the initial expectation of the study, which shows that in Vietnam, as cash flow increases, the level

of cash holdings also increases The coefficient of CashFlow variable is quite small, indicating that the effect of cash flow sensitivity on cash holdings is negligible However, this finding does not support the findings of previous studies, such as Riddick and Whited (2009), Bao, Chan and Zhang (2012) Riddick and Whited argued that an increase in the company's cash flow which indicated well-performing business, would lead to low level of cash stocking due to spending on tangible assets and new investment in profitable projects In order to further clarify the effect of cash-flow sensitivity on cash holdings in case of positive cash flows and negative cash flows, the study will estimate the main model of the study, added the dummy variable denotes negative cash flow (Neg) and the interactive variable CashFlow*Neg

Table 4 7: Results of the asymmetry sensitivity of cash flow

Hausman specification test between FEM and REM χ2 (df =9) = 28.96

(The sample includes non-financial companies from 2009 to 2015 The dependent variable ΔCashHoldings is the cash difference in year t and year t-1 divided by total assets, CashFlow is the income before extraordinary items Neg is the indicator variable of 1 if the company has negative cash flow and vice versa, Q is the total market value of capital plus the book value of the asset minus the book value The sum is the natural logarithm of the total asset, Expenditure is the capital expenditure divided by the total assets, Acquisition is the indicator variable of 1 if the company has the acquisition activity in year, ΔNCWC is the difference in non-cash flow divided by total assets, ShortDebt is short-term debt divided by total assets ***, **, * are respectively significant at 1%, 5%, 10%)