Why do vietnamese firms hold cash?

scipline of the board of directors or other outside parties. In addition, previous studies have shown that businesses that have excess money due to the self-interested behaviour of executives often squander money on inefficient projects. Accordingly, from the free cash flow perspective, only the factors of firm size, cash dividend policy, and investment opportunities are relevant to the firm's decision to hold cash. The remaining two theories, the trade-off theory and the pecking order theory, both assume that firms’ executives make decisions to maximize the interests of the firms, not their self-interests. However, from the trade-off theory perspective, the firm’s decision to hold cash is made to serve short-term transactions and for precautionary purposes. Therefore, factors such as long-term investment opportunities are not relevant. In contrast, the pecking order theory explains the firm's decision to hold cash on the basis of long term investment. Therefore, short-term considerations such as liquid asset substitutes, cash flow risks, or financial crisis risks are not relevant factors. In addition, because the pecking order theory does not project an optimal level of cash reserves as in the trade-off theory, it is not possible to make predictions about the impact of other internal funds (in this case, cash dividends) on the decision to hold cash. Nguyen Thanh Hong An and Hoang Mai Phuong 13 Companies can accumulate cash and other liquid asset substitutes at the same time to serve investment purposes in the future. Table 1 summarises the hypotheses about the factors that affect the amount of cash held by firms according to the three theories. Table 1. Theories and related hypotheses Firm characteristics Trade off theory Pecking order theory Free cash flow theory Future investment opportunities Positive Negative Cash flow Positive Positive Positive Cash flow uncertainty Positive Cash dividend policy Positive Negative Firm size Negative Negative Positive Leverage Positive Long-term debts Negative Profitability Positive Positive Positive Relationship with the bank Negative Negative Negative Liquid assets substitutes Negative 3. DATA DESCRIPTIONS To conduct the empirical analysis, we collected data on a sample of companies listed on the Ho Chi Minh City Stock Exchange (HOSE). Banking and financial services companies are excluded from the sample. The reason is that these companies have to comply with additional regulations on holding cash. Thus, their decisions to hold cash are not entirely based on economic considerations. Table 2. Sample structure by industry Industry Number of companies Number of observations Percentage Cumulative observations Cumulative percentage Wholesale 16 128 8.04 128 8.04 Retail 09 72 4.52 200 12.56 Information Technology 03 24 1.51 224 14.07 Accommodation and Catering 03 24 1.51 248 15.58 Mining 06 48 3.02 296 18.59 Production 76 608 38.19 904 56.78 Agriculture 04 32 2.01 936 58.79 Utilities 13 104 6.53 1,040 65.33 Transportation and Warehousing 16 128 8.04 1,168 73.37 Construction and Real Estate 53 424 26.63 1,592 100.00 Total 199 1,592 100.00 1,592 100.00 Notes: Companies are classified by their first registered type of business. Industries are classified in accordance with NAICS (North American Industry Classification System) 2007. DALAT UNIVERSITY JOURNAL OF SCIENCE [ECONOMICS AND MANAGEMENT] 14 In addition, we also chose companies listed from 2010 or earlier and that have financial information released continuously until the time of data collection to ensure the collection of adequate and reliable data, which help ensure the reliability of the statistical analysis results. As a result, 199 enterprises in 10 industries were selected for the survey, corresponding to 1,592 observations from 2011 to 2018. The relatively large number of observations and the structure of the sample covering businesses in many industries help ensure the sample representativeness. Table 2 presents information on sample structure, classified by major business activities. Tables 3 and 4 describe in detail the definitions and basic descriptive statistics of the variables used in this study. As in Opler et al. (1999) and Ferreira and Vilela (2004), we define the cash holding ratio as the amount of cash and cash equivalents divided by net assets, where net assets are book values of assets minus the amount of cash and cash equivalents. Table 3. Definition of research variables Variable Code Formula The ratio of cash holdings over net assets CASH Cash and cash equivalents Total assets - Cash and cash equivalents Market to book value MTB Total assets - Equity + Market value of equity Total assets The ratio of cash flows over net assets CFRATIO Net operating cash flows Total assets - Cash and cash equivalents Cash flows uncertainty CFUNCERTAINTY |CFRATIOit - CFRATIOi |̅̅ ̅̅ ̅̅ ̅̅ ̅̅ ̅̅ ̅̅ Cash dividend DIV Equal 1 if the company pay cash dividend and 0 otherwise Firm size LSIZE Log(Total assets) The ratio of total debts over net assets LEVERAGE Total debts Total assets - Cash and cash equivalents The ratio of long term debts over net assets MATURITY Long term debts Total assets - Cash and cash equivalents Profitability ROA Net profit after tax Total asset The ratio of bank borrowings over Total assets BORROWRATIO Bank borrowings Total assets The ratio of net working capital over net assets NWC Net working capital - Cash and cash equivalents Total assets - Cash and cash equivalents Nguyen Thanh Hong An and Hoang Mai Phuong 15 Table 4. Descriptive statistics Variable N Mean Median Std. dev. Min Max CASH 1,592 0.133731 0.059342 0.297952 0.000188 6.405646 MTB 1,592 1.288252 0.875000 3.601632 0.100000 121.340000 CFRATIO 1,592 0.071180 0.049837 0.164695 -0.749780 2.188948 CFUNCERTAINTY 1,592 0.081639 0.056605 0.097923 8.15E-06 1.199334 DIV 1,592 0.781407 1.000000 0.413422 0.000000 1.000000 LSIZE 1,592 9.163826 9.114875 0.536265 8.106911 11.459350 LEVERAGE 1,592 0.543082 0.555187 0.235398 0.030426 2.384434 MATURITY 1,592 0.138233 0.063652 0.199298 0.000000 4.490146 ROA 1,592 0.060375 0.047797 0.083371 -0.852590 0.783700 BORROWRATIO 1,592 0.253434 0.241600 0.189432 0.000000 0.975100 NWC 1,592 0.139682 0.115669 0.238880 -2.049630 0.925818 Notes: The statistics are calculated using 1592 observations from 199 companies listed on HOSE from 2011 to 2018; CASH is the ratio of cash and cash equivalents over net assets; MTB is the ratio of market to book value; CFUNCERTAINTY is the absolute difference between cash flow ratios for two adjacent periods; LSIZE is the logarithm of total assets; LEVERAGE is the ratio of total debts over net assets; MATURITY is the ratio of long term debts over net assets; ROA is the ratio of net profit over total assets; BORROWRATIO is the ratio of bank borrowings over total assets; and NWC is the ratio of net working capital over net assets. To estimate the impact of cash dividend policy, we define a dummy variable with a value of 1 if the company pays cash dividends in the observed year and 0 otherwise. To measure the performance of the company, we use the ratio of net profit after tax divided by the total value of the company's assets. Because the book value does not reflect the potential investment opportunities, investors often have to collect market information on companies’ investment opportunities and incorporate this information into stock prices. Therefore, we use the market value-to-book ratio to measure the company's potential investment opportunities. The larger the ratio, the more investment opportunities the company has and, accordingly, the higher the likelihood of the company's growth. To measure the value of highly liquid assets that can be sold when a business needs cash, we calculate the ratio of net working capital to net assets. Leverage is calculated by the ratio of total debt to net assets. The logarithm of a company's book value is used as a measure of its size. To measure the firm's ability to generate cash, we calculate the ratio of net cash flow from operating activities to the net assets of the corresponding year. Based on this variable, the cash flow variability of each firm is measured by taking the absolute value DALAT UNIVERSITY JOURNAL OF SCIENCE [ECONOMICS AND MANAGEMENT] 16 of the difference between the net cash flow from business activities on net assets and the average of this figure for the entire survey period. Long-term liabilities of the company are calculated by subtracting short-term liabilities divided by net assets. In addition, we calculate the bank loan ratio by dividing the business’s bank debt by its assets. 4. THE TRENDS OF CASH HOLDINGS OF VIETNAMESE FIRMS Referring to Table 5, the average ratio of cash to net assets of the companies in the research period is about 13.4%. This is lower than the average cash holding ratio of countries in the European Economic and Monetary Union (EMU) (14.8%) according to a study by Ferreira and Vilela (2004), or the corresponding US number (17.0%) according to research by Opler et al. (1999). The median value of 5.9% indicates that the typical cash holding ratio of firms in the sample is quite low and that high cash holdings occur only in a few companies. Compared to some Asian countries/territories, the median cash holding ratio of Vietnamese companies is higher than that of India (3.4%), Thailand (3.8%), and the Philippines (4.9%); but lower than Malaysia (6.3%), South Korea (8.9%), Indonesia (10.3%), Taiwan-R.O.C (11.6%), Hong Kong-P.R.C (13.1%), and Japan (15.5%) (Dittmar, Mahrt, & Servaes, 2003). Table 5. Cash holding ratio of Vietnamese firms in the period of 2011-2018 Year Mean Median Max Min Std. dev. Obs. 2011 0.133107 0.053759 1.092636 0.000987 0.185867 199 2012 0.145937 0.059293 5.238572 0.001605 0.398022 199 2013 0.150987 0.068639 3.513244 0.000467 0.292225 199 2014 0.165921 0.070360 6.405646 0.000188 0.488042 199 2015 0.152056 0.067763 2.280095 0.000576 0.289222 199 2016 0.118841 0.055353 2.171203 0.000769 0.210982 199 2017 0.112335 0.052354 2.634409 0.000564 0.223202 199 2018 0.090662 0.054060 0.899218 0.000788 0.120149 199 Total 0.133731 0.059342 6.405646 0.000188 0.297952 1,592 In addition, Table 5 also shows that the trend of holding cash increased gradually after 2011 (13.3%) and peaked in 2014 (16.6%) before continuously declining to 9.1% in 2018, with more than 50.0% of businesses having cash holdings less than 5.5%. Based on the descriptive statistics, we can infer that the cash holding ratio of Vietnamese companies is generally low and has tended to decrease in recent years. The increase in cash holding probably happens only in a few companies. If we look at the median over time, the common cash holding ratio of companies in the sample only fluctuates from 5.2% to 7.0% during the study period. Nguyen Thanh Hong An and Hoang Mai Phuong 17 Table 6 shows the difference in the amount of cash held among firms of different industries. The sector with the highest average cash holdings is Accommodation and Catering Services (86.0%), followed by Retail (25.5%) and Transportation and Warehousing (18.1%). The two industries with the lowest average cash holdings are Construction and Real Estate (7.4%) and Wholesale (10.0%). However, the averages are often strongly influenced by outlier observations. This makes the average trend measurement not accurately reflect the general trend, especially when the number of observations is small. This problem happens with a number of industries, such as Accommodation and Catering Services and Information Technology (with only three companies), Agricultural Production (with four companies), or Mining (with six companies). For a more comprehensive view, we refer to the median cash holding. According to this statistic, the industries with the highest proportion of cash holdings are Mining (12.6%) and Information Technology (12.5%). Meanwhile, Construction and Real Estate, Accommodation and Catering Services, and the Wholesale industry have the lowest proportion of cash holdings, with 3.5%, 3.7%, and 4.6%, respectively. Table 6. Cash holding ratio by industry Industry Mean Median Max Min Std. dev. Wholesale 0.099665 0.045674 0.551947 0.002829 0.122727 Retail 0.255419 0.093824 2.634409 0.008216 0.498662 Information Technology 0.145489 0.125234 0.579960 0.029647 0.118543 Accommodation and Catering 0.859166 0.037202 6.405646 0.003146 1.745637 Mining 0.143934 0.120256 0.530283 0.000188 0.125662 Production 0.126660 0.060063 1.136987 0.000446 0.171812 Agriculture 0.104090 0.088507 0.516208 0.002673 0.099536 Utilities 0.153989 0.068390 1.311618 0.000576 0.241806 Transportation and Warehousing 0.181198 0.099092 1.645856 0.004071 0.224938 Construction and Real Estate 0.073545 0.035171 0.917798 0.000904 0.098082 Total 0.133731 0.059342 6.405646 0.000188 0.297952 Notes: Companies are classified by their first registered area of business; Industries are classified in accordance with NAICS 2007. When classifying the companies into four quartile groups by size, the results show that companies in the first quantile (i.e., companies in the smallest 25.0% group) have the highest cash holding ratio, with an average holding ratio of 22.7% (median of 8.5%). Larger firms (in the second, third, and fourth quantiles) have lower average cash holding ratios, at 9.6% (median of 4.5%), 10.8% (median of 5.9%), and 10.3% (median of 5.3%). In addition, the decline in the proportion of cash held during 2016-2018 occurred in companies in all four quantiles but the most serious was in the group of small businesses (see Figure 1). DALAT UNIVERSITY JOURNAL OF SCIENCE [ECONOMICS AND MANAGEMENT] 18 .05 .10 .15 .20 .25 .30 .35 2011 2012 2013 2014 2015 2016 2017 2018 Quantile 1 Quantile 2 Quantile 3 Quantile 4 Figure 1. Cash holding ratio by firm size Referring to Table 7, firm performance seems to have a great impact on cash holding policy. Specifically, companies with positive profits often accumulate more than three times as much cash as loss-making companies (14.0% versus 4.0%). Similarly, enterprises with positive cash flows also accumulate twice as much cash as businesses with negative cash flows (16.0% compared to 8.0%). Table 7. Cash holding ratio by cash dividend policy, profitability, liquid asset substitutes, and cash flow Year Cash dividend Profit Liquid assets substitutes Cash flow Yes No Positive Negative Positive Negative Positive Negative 2011 0.14 0.06 0.14 0.02 0.11 0.19 0.17 0.07 2012 0.17 0.05 0.16 0.03 0.12 0.23 0.17 0.09 2013 0.17 0.07 0.16 0.05 0.15 0.16 0.17 0.12 2014 0.20 0.05 0.17 0.07 0.16 0.19 0.21 0.07 2015 0.19 0.05 0.16 0.05 0.15 0.17 0.18 0.09 2016 0.14 0.06 0.12 0.04 0.10 0.17 0.13 0.08 2017 0.13 0.05 0.11 0.07 0.10 0.14 0.13 0.08 2018 0.10 0.04 0.09 0.05 0.09 0.10 0.10 0.08 Mean 0.16 0.05 0.14 0.04 0.12 0.17 0.16 0.08 Nguyen Thanh Hong An and Hoang Mai Phuong 19 In addition, payment needs can also cause companies to hoard more cash. Businesses with negative liquid asset substitutes (measured by net working capital) often have to reserve more cash than businesses with positive liquid asset substitutes (17.0% versus 12.0%). Businesses that have pledged to pay cash dividends to their shareholders also have cash holding ratios three times higher than businesses that do not pay cash dividends (16.0% versus 5.0%). As shown in Table 8, we find that firms with better access to external capital often hold less cash. Specifically, firms with a high ratio of long term debt, bank debt, and leverage (both short-term and long-term), which fall into the third and fourth size quantiles, have cash holding ratios of 10.0%, 9.0%, and 12.0%, respectively, compared to 17.0%, 18.0%, and 15.0% for the group with little external borrowing (which fall into the first and second quartiles). Conversely, firms in the group with a lot of future investment opportunities or high cash flow risks often have a higher level of cash holding than businesses with few future investment opportunities or low cash flow risks (18.0% and 17.0% compared to 8.0% and 10.0%). Table 8. Cash ratio by future investment opportunities, long term debts, bank borrowings, cash flow uncertainty, and leverage Year MTB Long term debts Bank borrowings Cash flow uncertainty Leverage Q3-4 Q1-2 Q3-4 Q1-2 Q3-4 Q1-2 Q3-4 Q1-2 Q3-4 Q1-2 2011 0.19 0.11 0.08 0.18 0.08 0.19 0.16 0.09 0.11 0.16 2012 0.25 0.09 0.07 0.22 0.06 0.24 0.21 0.08 0.10 0.21 2013 0.22 0.09 0.09 0.21 0.10 0.21 0.17 0.13 0.13 0.17 2014 0.23 0.08 0.08 0.24 0.10 0.23 0.24 0.09 0.14 0.19 2015 0.20 0.09 0.13 0.18 0.12 0.19 0.17 0.14 0.16 0.14 2016 0.16 0.07 0.11 0.12 0.09 0.15 0.15 0.09 0.13 0.11 2017 0.14 0.07 0.11 0.12 0.09 0.14 0.12 0.11 0.11 0.11 2018 0.12 0.06 0.08 0.10 0.06 0.12 0.10 0.08 0.07 0.11 Mean 0.18 0.08 0.10 0.17 0.09 0.18 0.17 0.10 0.12 0.15 5. WHY DO VIETNAMESE FIRMS HOLD CASH? To answer this question, we perform three analysis steps. First, the correlation coefficient for the research variables is calculated to preliminarily evaluate the relationship between the ratio of cash holdings and certain characteristics of firms. Next, we conduct a regression analysis to measure and verify the relationship between the firm-specific variables and the firm's cash holding ratio, thereby identifying potential explanations for the cash holdings of Vietnamese firms. Finally, we conduct a robust test to reinforce the reliability of the conclusions drawn from the regression analysis. DALAT UNIVERSITY JOURNAL OF SCIENCE [ECONOMICS AND MANAGEMENT] 20 5.1. Correlation analysis The correlation coefficients and corresponding statistical significance levels are shown in Table 9. Specifically, the correlation coefficients between the cash holding ratio (CASH) and cash flow (CFRARIO) and the cash holding ratio (CASH) and firm performance (ROA) are positive and statistically significant, showing that firms with favourable conditions for accumulating cash often hoard more cash. In contrast, the negative and statistically significant correlation coefficients between the cash holding ratio (CASH) and net working capital (NWC), the cash holding ratio (CASH) and the ratio of bank loans (BORROWRATIO), the cash holding ratio (CASH) and the ratio of long term borrowings (MATURITY), and the cash holding ratio (CASH) and the size of the firm (LSIZE) show that firms with abundant cash replacement resources or those in a good position to access external capital often have lower amounts of cash. In addition, the positive correlation coefficients between cash ratio (CASH) and the change of cash flow (CFUNCERTAINTY) and between cash ratio (CASH) and the leverage ratio (LEVERAGE) indicate that companies with higher levels of financial risk often hold more cash. Finally, Table 9 also shows that firms paying cash dividends often hold a higher amount of cash than firms that do not pay dividends, and the firm's investment prospects do not seem to be related to the amount of cash held. Table 9. Correlation coefficients between research variables CASH MTB ROA CFRATIO CFUN- CERTAINTY DIV CASH 1.000000 MTB 0.018134 1.000000 ROA 0.309230*** 0.064337** 1.000000 CFRATIO 0.536636*** 0.021469 0.349558*** 1.000000 CFUNCERTAINTY 0.371144*** -0.006400 0.045956* 0.347767*** 1.000000 DIV 0.143828*** -0.050720** 0.299470*** 0.122890*** -0.022010 1.000000 LSIZE -0.139780*** 0.094052*** -0.058580** -0.129810*** -0.180540*** 0.065471*** LEVERAGE 0.093301*** 0.014309 -0.364080*** -0.140770*** 0.045740* -0.034040 MATURITY -0.063880** -0.031240 -0.152400*** -0.023890 -0.134850*** 0.018585 BORROWRATIO -0.212970*** -0.000890 -0.387700*** -0.248590*** -0.032160 -0.104450*** NWC -0.205230*** 0.014549 0.210194*** -0.071380*** -0.020350 -0.021040 LSIZE LEVERAGE MATURITY BORROWRATIO NWC LSIZE 1.000000 LEVERAGE 0.229393*** 1.000000 MATURITY 0.244207*** 0.294988*** 1.000000 BORROWRATIO 0.272129*** 0.568452*** 0.255953*** 1.000000 NWC -0.067950*** -0.552140*** -0.081990*** -0.375470*** 1.000000 Notes: *, **, and *** correspond to the 10%, 5%, and 1% levels of significance, respectively. Nguyen Thanh Hong An and Hoang Mai Phuong 21 The results of the above correlation analysis seem to show that Vietnamese companies hold money primarily for transactional and precautionary purposes rather than for long term investment or for the benefit of the executives. They increase cash holdings when business conditions are favourable, but also consider cash substitutes, such as net working capital or bank loans. This view is underpinned by the very low average cash holding ratio of most companies and the decreasing trend of cash holdings in recent years. While the correlation analysis reveals much interesting information, the result is not conclusive. The reason is because correlation analysis only describes the relationship between two variables without taking into account the interaction between them and other variables. Furthermore, correlation analysis does not take into account the causality of the relationships. Ignoring these two characteristics may cause the interpretation of the statistical analysis results to be misleading. In order to have a more comprehensive and accurate view of the impact of factors on the company's cash holding ratio, regression analysis is performed in the next section. 5.2. Regression analysis To analyze the impact of the research variables on the ratio of cash holding, we perform a regression analysis as follows. First, the usual regression model with the pooled data (denoted as POLS (Pooled Ordinary Least Squares)) is estimated: CASHit = β1 + β2MTBit + β3CFRATIOit + β4CFUNCERTAINTYit + β5DIVit + β 6 LSIZEit + β7LEVERAGEit + β8MATURITYit + β9ROAit (1) + β 10 BORROWRATIOit + β11NWCit + εit However, the ordinary least squares regression model with pooled data can produce inconsistent results as it ignores the impact of unobserved factors at the firm level. The estimation results of Model (1) are presented in Column A of Table 10. To compare and select a more effective model, we modify the structure of Model (1) to include unobserved factors at the firm level: CASHit = β1+ β2MTBit + β3CFRATIOit + β4CFUNCERTAINTYit + β5DIVit + β 6 LSIZEit + β7LEVERAGEit + β8MATURITYit + β9ROAit (2) + β 10 BORROWRATIOit + β11NWCit + μi + εit Model (2) is estimated by two methods. First, Model (2) is estimated using the GLS (Generalized Least Squares) method with the assumption that the unobserved factors at the firm level are random with zero average and are independent of the explanatory variables in the model. This model is denoted as REM (Random Effects Model) and the estimation results are presented in Column B of Table 10. Model (2) is re-estimated with the assumption that the unobserved firm-level factors are not random and possibly correlated with the explanatory variables in the model. This model is DALAT UNIVERSITY JOURNAL OF SCIENCE [ECONOMICS AND MANAGEMENT] 22 denoted as FEM (Fixed Effects Model) and the estimation results are presented in Column C of Table 10 To choose between the three models (POLS, REM, and FEM), we perform the Breusch-Pagan LM (Lagrange multiplier) test to choose between the POLS model and the REM model and the Hausman test to choose between the REM model and the FEM model. One of the issues that can affect the test results is the distribution properties of the error term of Model (2). Previous studies have also shown that because the proportion of cash held is always positive, the regression results of Model (2) can produce errors that have a distribution which is different from the normal distribution, thereby affecting the validity of the tests (Bates et al., 2009). To remedy, we regress Model (2) using the logarithm of CASH (denoted LCASH) as the dependent variable. This approach is deemed to help the estimates achieve distribution properties that are closer to the normal distribution. Therefore, we estimate Model (3) and the estimation results are presented in Column D of Table 10: LCASHit = β1+ β2MTBit + β3CFRATIOit + β4CFUNCERTAINTYit + β5DIVit + β 6 LSIZEit + β7LEVERAGEit + β8MATURITYit + β9ROAit (3) + β 10 BORROWRATIOit + β11NWCit + μi + εit To account for the change in the amount of cash holdings over the years, dummy variables that encapsulate the impact of the time factor are included in the structure of the three models (not presented in the formulae). In addition, the estimated standard errors of the models are adjusted for heterogeneity and serial correlation using the method presented in Arellano (1987). To test for the existence of a multi-collinearity problem, the variance inflation factors (VIF)’s of the independent variables in the models are calculated. The results (not presented) show that the average of the VIF’s is 1.57 and no VIF of any variable exceeds the value of 3. Referring to Table 9, the correlation coefficients between the independent variables are also very low. The statistical evidence mentioned above shows that multi-collinearity is not a serious problem. Based on the results of Breusch-Pagan LM test from Table 10, the hypothesis that the POLS model is a suitable model is rejected. Between the other two models, the Hausman test shows that the FEM model structure is, indeed, a more suitable one. Therefore, we use the FEM model to perform the next analysis steps to figure out the factors that affect the amount of cash held by the firms. Based on the FEM model, we find that the regression coefficients of the CFRATIO and ROA variables are both positive and statistically significant at 1%. This shows that businesses tend to hold more cash when business conditions are favourable. The regression coefficients of the BORROWRATIO, LSIZE, and NWC variables are all negative and statistically significant at 5% or higher, suggesting that companies often Nguyen Thanh Hong An and Hoang Mai Phuong 23 reduce the amount of cash held when other sources of cash substitution (excess working capital or loans from banks) become more abundant or more accessible. Table 10. Regression results Variable POLS (A) REM (B) FEM (C) Log(Cash) (D) C 0.157399 0.354104** 0.451913** 0.2088445* MTB 0.000023 -0.000164 0.000266 -0.020206*** CFRATIO 0.631320*** 0.5335301*** 0.472683*** 0.732942*** CFUNCERTAINTY 0.634263*** 0.305838*** 0.190045 0.321758 DIV 0.022486 0.027349** 0.020857* 0.156602*** LSIZE -0.026103** -0.051097*** -0.067221** -0.170040 LEVERAGE 0.335456*** 0.448380*** 0.579273*** 2.175882*** MATURITY -0.026489 0.004876 -0.005535 -0.142808 ROA 0.728976*** 0.600218*** 0.477420*** 1.918083*** BORROWRATIO -0.367070*** -0.337246*** -0.320349*** -2.6611105*** NWC -0.202528*** -0.223244*** -0.221754** -0.340727* Year dummies Yes Yes Yes Yes N 1,592 1,592 1,592 1,592 R2 0.452096 0.365879 0.733057 0.745819 F 76.39791 53.422020 17.57516 18.77891 Prob(F) 0 0 0 0 Breusch-Pagan LM 1245.090000 Prob(LM Chi-sqr) 0 Hausman Chi-sqr 110.685317 Prob(Chi-sqr) 0 Notes: *, **, and *** correspond to the10%, 5%, and 1%levels of significance, respectively; The statistics are calculated using 1592 observations from 199 companies listed on HOSE from 2011 to 2018; CASH is the ratio of cash and cash equivalents over net assets; MTB is the ratio of market to book value; CFUNCERTAINTY is the absolute difference between cash flow ratios of the two adjacent periods; LSIZE is the logarithm of total assets; LEVERAGE is the ratio of total debts over net assets; MATURITY is the ratio of long term debts over net assets; ROA is the ratio of net profit over total assets; BORROWRATIO is the ratio of bank borrowings over total assets; and NWC is the ratio of net working capital over net assets. DALAT UNIVERSITY JOURNAL OF SCIENCE [ECONOMICS AND MANAGEMENT] 24 On the other hand, the regression coefficient of the LEVERAGE variable has a positive sign, indicating that the more debt a firm has, or the higher its financial risk, the more cash it holds. This result is inconsistent with the results of Bates et al. (2009) for the US market or research results of Chen, Dou, Rhee, Truong, and Veeraraghavan (2015) for a multinational dataset. Unlike the initial observations in the correlation analysis, the CFUNCERTAINTY variable, another variable representing the volatility of financial ability, does not seem to have a significant effect on the business decision to hold cash. In the same manner, the regression coefficients associated with the MTB variable are not statistically significant, implying that potential future investment opportunities do not matter in the decision-making process about the amount of cash held. Finally, the regression coefficient associated with the DIV variable is only statistically significant at 10%, suggesting that the dividend policy does not seem to have an impact on the decision to hold cash. Reconciling the results of the regression model with CASH as the dependent variable (Column C of Table 10) with the regression model using LCASH as the dependent variable (Column D of Table 10), we can see some noticeable differences. The MTB variable is now statistically significant at 1% and negatively correlated with the dependent variable, meaning that firms with lower prospects of future investment seem to accumulate more cash. This is consistent with the predictions of the free cash flow theory (Ferreira & Vilela, 2004). In contrast, the LSIZE variable is no longer statistically significant, indicating that firm size has no effect on the decision to hold cash. This result is inconsistent with the results from the US market, where large companies often hold less cash (Bates et al., 2009), or evidence from research by Dittmar et al. (2003)for a multinational dataset. In addition, the DIV variable (cash dividend policy), which is not statistically significant in Model (3), is now statistically significant at 1% in Model (4), implying that companies paying cash dividends tend to hold more cash. This result is inconsistent with Almeida et al. (2004), Ferreira and Vilela (2004), and Opler et al. (1999). In these studies, the researchers argue that firms paying dividends are often considered to have less difficulty accessing external capital and, as a result, there is no need to hoard much cash. Normally, dividend policy is usually kept stable for a long period (Lintner, 1956). Firms are very reluctant to change dividend policy, especially to reduce dividends, because this action may convey negative information about the performance of the company (Miller & Modigliani, 1961). Therefore, for firms that have a cash dividend payment policy, it is likely that they consider this as a spending item that needs to be planned in advance. This statistical result seems to support the trade-off theory of the company's decision to hold cash. Taking the above results together, we argue that the decision to hold cash is a calculated decision, based on the consideration of costs and benefits to the business. This behaviour is in line with the prediction of the trade-off theory. However, there is also evidence that companies hoard cash for the benefit of the executives. Specifically, the results show that companies with less investment opportunities in the future often accumulate more cash than those with more investment opportunities. The act of Nguyen Thanh Hong An and Hoang Mai Phuong 25 hoarding cash in this situation seems to indicate that firms with less investment opportunities hold cash to serve the interests of the executive rather than the interests of the shareholders. The simultaneous existence of evidence supporting the three theories on the purposes of holding cash is not surprising. Theoretically, the three most common theories about corporate motivations to hold cash have many overlapping predictions (Opler et al., 1999). In fact, a business can still determine the amount of cash held to serve a variety of purposes. To determine which cash holding motivation is stronger, we carry out the robust test. 5.3. Robust tests A prediction specific to the trade-off theory that differentiates it from the other two theories is that it predicts an optimal level of cash holdings (Opler et al., 1999; Orlova & Rao, 2018). When the amount of cash holdings is considered too high, companies tend to decrease holdings in the next period. Likewise, when cash holdings are lower than desired, companies tend to increase holdings in the next period. This results in a negative correlation between the change in current cash holding ratio (CASHt) and changes in the cash holding ratio of the previous period (CASHt-1) (Opler et al., 1999). Thus, if the conclusion in the regression analysis section is correct, i.e., that companies consider the benefits and costs when deciding on the ratio of cash held, we expect to observe a significant negative regression coefficient between the two variables, CASHt and CASHt-1, as specified by Model (4): ∆CASHit = β1 + β2∆CASHit-1+ μi + εit (4) Technically, we estimate Model (4) using two methods. For the first method, we perform regression Model (4) for each company. A new data set is formed with 199 observations, consisting of the regression coefficients between CASHt and CASHt-1 for each enterprise. Then, a univariate test is performed to decide whether the average regression coefficient between CASHt and CASHt-1 of the companies is significantly negative or not. This method is also used in Opler et al. (1999). The results of this method are presented in Figure 2. We can see that the average regression coefficient of 199 companies in the sample is -0.302576 (with standard deviation of 0.535707) and the median is -0.298583. Performing a univariate test to decide whether the average regression coefficient is actually negative or not, we have the test statistic t = -7.967714 (prob(t) = 0.0000), indicating that the mean of the slope coefficients is negative and statistically significant at the 1% level. Thus, we can conclude that the average regression coefficient between CASHt and CASHt-1 of the companies is negative and statistically significant. This shows that the business has adjusted the holding rate of cash to the desired level, which is consistent with the forecast of the trade-off theory. DALAT UNIVERSITY JOURNAL OF SCIENCE [ECONOMICS AND MANAGEMENT] 26 0 10 20 30 40 50 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 Figure 2. Histogram of the regression coefficients between CASHt and CASHt-1 Notes: The regression coefficient of each individual company is estimated using the following model: ∆CASHt = β1+ β2∆CASHt-1+ εt. However, this method also has the limitation that there are only eight observations (over eight years) for each company. Thus, the estimation of Model (4) at the company level may yield inaccurate results. To achieve more reliable results, we regress Model (4) with panel data. Although the results are more reliable due to the larger number of observations, this estimate is still biased (unrealistically small) and may be inconsistent due to the endogeneity problem when the lag dependent variable is used as an explanatory variable (Nickell, 1981). To remedy, we use the Dynamic Generalised Method of Moment (DGMM) as presented in Arellano and Bond (1991) to estimate the regression coefficients. Regression results using these two techniques are presented in Table 11. Table 11. Regression results between CASHt and CASHt-1 using DGMM Variable POLS FEM Dynamic GMM C -0.010395** -0.010452*** - CASHt-1 -0.341588*** -0.358129*** -0.305148*** Year dummies Yes Yes Yes N 1,194 1,194 995 R2 0.155357 0.235195 F 38.387850 1.490881 Prob(F) 0 0 Hansen J 19.623750 Prob(J) 0.142457 AR(2) -0.874611 Prob(AR2) 0.381800 Notes: *, **, and *** correspond to the 10%, 5%, and 1%levels of significance, respectively. Nguyen Thanh Hong An and Hoang Mai Phuong 27 According to the results in Table 11, we find that the regression coefficient corresponding to the variable CASHt-1 is negative and statistically significant at 1% by all three estimation methods. As expected, the method of estimating POLS and FEM ignores endogenous factors in Model (4), so the impact of CASHt-1 is estimated lower than reality (-0.340 and -0.350 compared with -0.305). The regression coefficient of the CASHt-1 variable has a negative sign, indicating that the enterprise adjusts the amount of cash holdings in the current period downward when the amount of money in the previous period is too high. Conversely, companies would increase their cash holdings in the present period if the amount of cash holdings in the previous period is too low. The absolute value of 0.30 for the regression coefficient also indicates that the change in the cash ratio of the following year is about 30% of the previous year. This also means that the rate of adjustment of the holding rate of the business is about 30% per year, which is higher than the corresponding value of 24.2% for the US market as estimated by Opler et al. (1999). The regression results with the FEM model and the robust tests support the trade-off theory, implying that listed companies in Vietnam hold cash mainly for transactional and precautionary purposes. This conclusion is also consistent with the recent research results of Chen et al. (2015) for a dataset of businesses from 72 countries. Moreover, the cash holdings of Vietnamese firms are generally low and have been on a downward trend in recent years. This may be a bad signal for the competitiveness and growth of Vietnamese firms in the future. The reason is that, according to the pecking order theory, cash is also used as an additional capital source for corporate investments, especially investments in R&D. Investments in R&D are generally hard to finance by external sources of capital. The fact that Vietnamese enterprises hold too little cash is an indication of low internal investment capacity. This can lead to a lack of investment in research and development. To the extent that this is true, the growth and competitiveness of businesses in the future may be negatively affected. Last, but not least, the regression results also show that businesses with less investment and growth opportunities in the future (proxied by low MTB values) tend to increase cash holdings. According to free cash flow theory, this is a negative sign. In particular, the executives of these companies may have decided to hoard cash to serve their own interests rather than the shareholders’ and the board of directors has not fulfilled their monitoring and disciplining responsibilities. 6. CONCLUSIONS Based on statistical analysis, this study has shown that the amount of cash held by Vietnamese firms in the period of 2011-2018 is quite low (an average of about 13.4% of net assets), with more than 50.0% of enterprises holding less than 5.9% of net assets. This value is low compared to businesses in western countries, such as 14.8% of businesses in EMU countries (Ferreira & Vilela, 2004) or 17.0% of US businesses in the period before 1999 (Opler et al., 1999) or more than 20% on average between 2000 and 2006 (Bates et al., 2009). Another noteworthy point is that the average amount of cash held by Vietnamese companies has been on a downward trend in the last three DALAT UNIVERSITY JOURNAL OF SCIENCE [ECONOMICS AND MANAGEMENT] 28 years, with the average amount of cash held decreasing from a peak of 16.6% of net assets in 2014 to below 9.1% of net assets in 2018 (with more than 50.0% of businesses holding less than 5.4% of net assets). The hoarding of large amounts of cash only occurs in a few specific businesses and is not a common feature of listed companies in Vietnam. Results of regression analysis between the ratio of cash held and a number of business characteristics show that it is likely that Vietnamese firms make decisions on the amount of cash held based on cost and benefit considerations and the main motivation of holding cash is probably to serve transactions and prevent short-term business volatility rather than to accumulate long-term investment or to serve the self- interest of the executives. Analysis of cash adjustments shows that firms often increase the amount of cash held in the next period when the amount of cash in the previous period is lower than desired and vice versa. This is an additional evidence supporting the trade-off theory of corporate cash holdings in the context of Vietnam. The research results have two implications for researchers and investors. Firstly, the analyses show that for certain businesses, namely firms with low growth potential and limited investment opportunities in the future, managers may have made decisions regarding cash holdings to serve their own self-interest rather than the shareholders’. This also implies that the board of directors of these firms may have failed to fulfil their over-sight and disciplinary functions regarding the behaviour of executives toward the best interest of the shareholders. Secondly, the fact that Vietnamese enterprises have a relatively low cash holding rate that mainly serves transactional and precautionary purposes could be a negative sign. It shows that the internal investment capacity of Vietnamese firms is low. This may limit their ability to invest in Research and Development (R&D), which, in turn, negatively affects the long-term growth and competitiveness of Vietnamese firms. While making contributions to the literature of corporate cash holding decisions, this study cannot avoid limitations. Firstly, this study is limited to the period from 2011 to 2018 and to companies listed for at least nine years on the Ho Chi Minh City Stock Exchange. Further studies may expand the scope of research over a longer period of time and include businesses listed on other stock exchanges in Vietnam. This would help reveal long-term trends and provide more evidence on Vietnamese firms' decisions to hold cash. Secondly, this study shows that Vietnamese firms have reduced their cash holdings in recent years. However, the reasons for this downward trend are still unaccounted for. A number of studies in other countries have shown that changes in the macro background may cause changes in corporate cash holdings over time (Almeida et al., 2004). Therefore, follow-up studies should focus on discovering the causes of this phenomenon to complete the picture of Vietnamese firms’ decisions to hold cash. Thirdly, this study finds that the board of directors of firms, especially firms with low growth and few investment opportunities, may have failed to monitor and discipline the executives to the best interest of the shareholders. In order to have a more specific view, further studies may analyse the impact of corporate governance quality or ownership structure on the cash holding decisions of Vietnamese firms. Lastly, and most Nguyen Thanh Hong An and Hoang Mai Phuong 29 importantly, this study shows that Vietnamese firms hold relatively less cash than firms in other countries. This may be a sign of low internal investment capacity. If so, the long-term growth and competitiveness of Vietnamese firms may be negatively affected. For more conclusive evidence, future studies need to investigate the impact of cash holdings on investment spending (especially investments in R&D) of Vietnamese firms. ACKNOWLEDGEMENT This research is funded by a Dalat University research grant. REFERENCES Acharya, V. V., Almeida, H., & Campello, M. (2007). Is cash negative debt? A hedging perspective on corporate financial policies. Journal of Financial Intermediation, 16(4), 515-554. Almeida, H., Campello, M., & Weisbach, M. S. (2004). The cash flow sensitivity of cash. Journal of Finance, 59(4), 1777-1803. Arellano, M. (1987). Computing robust standard errors for within-groups estimators. Oxford Bulletin of Economics and Statistics, 49(4), 431-434. Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. The Review of Economic Studies, 58(2), 277-297. Bates, T. W., Kahle, K. M., & Stulz, R. M. (2009). Why do U.S. firms hold so much more cash than they used to? Journal of Finance, 64(5), 1985-2021. Baumol, W. J. (1952). The transactions demand for cash: An inventory theoretic approach. The Quarterly Journal of Economics, 66(4), 545-556. Chen, Y., Dou, P. Y., Rhee, S. G., Truong, C., & Veeraraghavan, M. (2015). National culture and corporate cash holding around the world. Journal of Banking and Finance, 50(1), 1-18. Dittmar, A., Mahrt, S. J., & Servaes, H. (2003). International corporate governance and corporate cash holdings. Journal of Financial and Quantitative Analysis, 38(1), 111-133. Faulkender, M. (2002). Cash holdings among small businesses. Retrieved from https://papers.ssrn.com/sol3/papers.cfm?abstract_id=305179. Fazzari, S. M., Hubbard, G. R., & Petersen, B. C. (1988). Financial constraints and corporate investment. Brooking Papers on Economic Activities, 1988(1), 141-206. Ferreira, M. A., & Vilela, A. S. (2004). Why do firms hold cash? Evidence from EMU countries. European Financial Management, 10(2), 295-319. Harford, J. (1999). Corporate cash reserves and acquisitions. Journal of Finance, 54(6), 1969-1997. DALAT UNIVERSITY JOURNAL OF SCIENCE [ECONOMICS AND MANAGEMENT] 30 Harford, J., Mansi, S. A., & Maxwell, W. F. (2008). Corporate governance and firm cash holdings in the US. Journal of Financial Economics, 87(3), 535-555. Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance and takovers. The American Economic Review, 76(2), 323-329. Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4), 305-360. Keynes, J. M. (1936). The general theory of employment, interest, and money. London, UK: Macmillan and Co. Limited Publishing. Kinh, K. (2018). "Đau đầu" vì nhiều tiền: Chân dung những doanh nghiệp đang nắm giữ lượng tiền mặt lên tới cả chục nghìn tỷ. Retrieved from dau-vi-nhieu-tien-chan-dung-nhung-doanh-nghiep-dang-nam-giu-luong-tien- mat-len-toi-ca-chuc-nghin-ty-20180220093509968.chn. Lintner, J. (1956). Distribution of incomes of corporations among dividends, retained earnings, and taxes. The American Economic Review, 46(2), 97-113. Malmendier, U., & Tate, G. (2005). CEO overconfidence and corporate investment. Journal of Finance, 60(6), 2661-2700. Miller, M. H., & Modigliani, F. (1961). Dividend policy, growth, and the valuation of shares. The Journal of Business, 34(4), 411-433. Miller, M. H., & Orr, D. (1966). A model of the demand for money by firms. The Quarterly Journal of Economics, 80(3), 413-435. Modigliani, F., & Miller, M. H. (1958). The cost of capital, corporate finance, and the theory of investment. The American Economic Review, 48(3), 261-297. Myers, S., & Majluf, N. (1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics, 13(2), 187-221. Nguyen, T. H. A., & Nguyen, V. T. (2018a). Free cash flow and corporate profitability in emerging economies: Empirical evidence from Vietnam. Economics Bulletin, 38(1), 211-220. Nguyen, T. H. A., & Nguyen, V. T. (2018b). Working capital management and corporate profitability: Empirical evidence from Vietnam. Foundations of Management, 10, 195-206. Nguyễn, T. U. U., & Từ, T. K. T. (2015). Ảnh hưởng của việc nắm giữ tiền mặt vượt trội đến các quyết định tài chính của các doanh nghiệp Việt Nam. Tạp chí Phát triển và Hội nhập, 25(35), 36-45. Nguyễn, T. U. U., & Từ, T. K. T. (2017). Ảnh hưởng của dòng tiền đến độ nhạy cảm tiền mặt nắm giữ trong điều kiện hạn chế tài chính của các công ty Việt Nam. Tạp chí Phát triển Kinh tế, 28(11), 26-53. Nickell, S. (1981). Biases in dynamic models with fixed effects. Econometrica, 49(6), 1417-1426. Nguyen Thanh Hong An and Hoang Mai Phuong 31 Okzan, A., & Okzan, N. (2004). Corporate cash holdings: An empirical investigation of UK companies. Journal of Banking and Finance, 28(9), 2103-2134. Opler, T., Pinkowitz, L., Stulz, R. M., & Williamson, R. (1999). The determinants and implications of corporate cash holdings. Journal of Financial Economics, 52(1), 3-46. Orlova, S. V., & Rao, R. P. (2018). Cash holdings speed of adjustment. International Review of Economics and Finance, 54(C), 1-14. Phạm, H., & Đinh, P. Q. T. (2018). Ảnh hưởng hạn chế tài chính đến quyết định nắm giữ tiền mặt của các doanh nghiệp niêm yết tại Việt Nam. Journal of Science Ho Chi Minh City Open University, 60(3), 133-143.