Luận án Thông tin tài chính tác động đến suất sinh lời chứng khoán của các công ty niêm yết tại thị trường chứng khoán Việt Nam

t Thomson Reuter Pt - Pt - 1 Pt – 1 (Breyley & Myers (2004), Nguyen Anh Phong (2015)) 15 SUE Thomson Reuter Eit – Eit - 1 σit (Chordia (2006), Vinh & Phuong (2014)) DVOL Thomson Reuter LN(Pit * Volumeit) (Chordia (2006), Vinh & Phuong (2014)) SIZE Thomson Reuter LN(total assets) (Banz (1981), Chordia (2006, 2014), Vinh & Phuong (2014)) BM Thomson Reuter = (Total assets – Debt – Fixed intangible assets) / ( No. of Oustanding shares * Stock price) (Weiqiang (2008); Chordia (2006); Vinh & Phuong (2014); Fama & French (1992)). RET12 Thomson Reuter Pit-2 – Pit - 12 Pit – 12 (Jegadeesh (1993); Chordia (2006); Vinh &Phuong (2014)) |DA| Thomson Reuter |DAi ,t| = TAi, t / Doanh thui,t –TAi,t-1/ Doanh thu i,t-1 (Friedlan (1994); DeAngelo (1986)) TA (Total accruals) = LNST - Cash flow from operations Audit-S Auditor size ( KPMG, EY, PwC, Deloitte, AASC, A&C: dummy = 1 (auditor belongs to Big6) and 0 for other auditing firms. (Ranking criteria of auditing firms 2015 AASC) Source: Author‘s compilation 3.2. Research data The study focuses on the period from 01/2010 and 06/2016, of 482 firms that have continuous listing time on HOSE and HNX for at least 2 years. Firms that have book value < 0 will be eliminated. The data are collected from webstites of State Stock Commission, and www.cophieu68.vn and database of Thomson Reuters. 3.3. Research method 3.3.1. Impact of the time of financial statement publication on stock return (Model 1) 16 Event study: the length of windows in this study: [-90;90], [-5;5]. The following timeline defines how these windows are used: Graph 3.1: Length of windows Estimation window Event window Post - event window (1) Good news, bad news are defined based on UE (Bajo, 2010): good news is when UE >0 and bad news UE <0. According to Begley, Joy and Paul Fischer 1998 UE can be obtained from the following formula: 1,,, EPS  tititi EPSUE (16) EPS is calculated following Stock exchange: EPS = Profit/Loss allocated to common shareholders Average outstanding shares in the period (17) SSL (Rt): according to Breyley and Myers (2004) and Nguyen Anh Phong (2015) with the data from Thomson Reuters: Rt = Pt - Pt - 1 + Dt Pt - 1 Trong đó: Rt is the return at t; Pt is the stock price at t Pt-1 is the stock price at (t-1) ; Dt is the stock dividend (if any) at t (18) Market return (Rm) Rm is the market return, this variable is based on VNIndex and HNXindex. Based on the daily close price, this variable is calculated using the following formula: - For firms listed on HOSE : Rm,t = Vnindexi,t – Vnindexi,t-1 Vnindexi,t-1 - For firms listed on HNX: Rm,t = HNX-Indexi,t – HNX-Indexi,t-1 HNX-Indexi,t-1 (4) Abnormal return (AR): ARi,t = Ri,t – αit– βRm Where: - ARit is the abnormal return of stock i at t - Ri,t is the stock return of stock i at t 17 - Rm : market return at t - it : constant ;  is systematic risk (19) Cumulated average abnormal return (CAAR): CAAR is the sum of the differences between expected return from a stock and actual return around the publication date of financial statements: Where : AARt is the abnormal average return, calculated as the average of abnormal returns for all firms in the sample from daily stock prices, with ARi, t+k being the abnormal return of stock i at time t + k; is the cumulated stock return of stock i from t to t + k. Sample variance is calculated as follows : Where : : average of AAR in 181 transaction days, without the effect of income information CAR is expected to follow normal distribution: (5) The author sorts CAAR of firms following the windows: [-5, - 4, -3,- 2, 0, 1, 2, 3, 4, 5 ] for short-term windows, and in long-term windows [-90, -60, -30, -10, 0, 10, 30, 60, 90 ], and divided into 2 groups: UE > 0 and UE < 0, in order to test whether CAAR changes depending on UE > 0, UE<0 for the above windows. The results will help conclude if the time of financial statement publication has impact on stock returns and whether good news and bad news can influence CAAR. (6) Conduct t test to check if CAAR is different from 0 significantly in each case (7) The thesis examines each industry: Basic Materials, Consumer Goods, Consumer Services, Health Care, Industrials, Oil & Gas, Technology, Others. For firms that publish financial statements on time (Bagloni, 2002): )*(CAAR ,,7,4,10, tk negUE tk negUE tktktk UEDDUE   For firms with late publication (Bagloni, 2002) 18 )**(）*( )*()*(CAAR ,,,10,,8 ,,7,,5,4,2,10, tk negUE tk late tk negUE tk late tk ti negUE tktk late tk negUE tk late tktktk UEDDDD UEDUEDDDUE     For firms with early publication (Bagloni, 2002) )**()*()*( )*(CAAR ,,,11,,9,,7 ,,6,4,3,10, ti negUE tk early tk negUE tk early tktk negUE tk tk early tk negUE tk early tktktk UEDDDDUED UEDDDUE     where: CAARkt is the abnormal cumulated average return (Bagnoli, 2002) of firms in industry k at time t; D late is a dummy indicating firm issues financial statements late; Dearly is a dummy indicating firm issues financial statements early; NegUE is a dummy variable, indicating that unexpected earnings of firm is negative; UE is the difference between actual and expected abnormal income on each share (Begley, Joy and Paul Fischer, 1998; Bajo, 2010). K industries include: Basic Materials, Consumer Goods, Consumer Services, Health Care, Industrials, Oil & Gas, Technology, Others. (8) The author estimate models to examine the impact of the time of financial statement publication on abnormal stock return, using Pooled OLS. The thesis further provides descriptive statistics, correlation analysis and Woolridge test and Wald test for unbalanced panel data. The author will propose ways to overcome problems related to violations of OLS assumptions (if any) as suggested by tests. 3.3.2. Research methodology for the model on the impact of information contained on financial statements and stock returns In model 2, the study uses unbalanced panel data and Stata 12.0 to process data. The study employs 3 methods: Pooled OLS, Fixed effects model (FEM) and Random effects model (REM). Tests for model selection: Selection between Pooled OLS and FEM: the study employs F test to decide the choice between Pooled OLS and FEM. Choice between FEM and REM: the study employs Hausman test to decide. Wald test is used to test autocorrelation; Woolridge test is used to test for the heteroskedasticity. If tests suggest there is either heteroskedasticity or autocorrelation, the author will overcome by using robust/cluster option. CONCLUSION OF CHAPTER 3 Chapter 3 presents in detail the methodology and data description for the study to deal with research objectives. 19 CHAPTER 4: RESEARCH RESULTS AND DISCUSSION 4.1. Analysis of current status 4.1.1. The current status on the regulations regarding information publication in Vietnam From 2000 till now, the regulatory papers about information publication in Vietnamese stock market have had great changes to accommodate market‘s revolution. Circular No. 48/1998/ND-CP is the first paper from the government providing guidance on information publication. Next is Circular 144/2003/ND-CP, Law on Stock 70/2006/QH11, Circular 38/2007/TT-BTC, Circular 09/2010/TT-BTC in replacement of Circular 38/2007/ND-CP, Circular 85/2010/ND-CP in replacement of Circular 36/2007/ND-CP, being the prerequisite to complementing part of Law on stock, but the levels of fine imposed are not strong enough to act as a deterrent to violations. In 2012, Circular 52/2012/TT-BTC was issued, replacing Circular no. 09/2010/TT-BTC with many new regulations; Circular 155/2015/TT-BTC in replacement of circular 52/2012/TT-BTC with many changes. Moreover, Circular 155 also requires English version of information to be published on Stock Exchanges and Stock Depository Centers. State Stock Commission may grant an extension to the period of financial statement publication, but not more than 100 days, from the end of fiscal year upon receiving written text from firms. In cases of abnormal publication needed, in order to allow for the timeliness of the information, Circular 155 states that the allowed time for publication is 24 hours, which is much shorter than the 72 hours allowed in Circular 52. Circular 155 also stipulates the same extension to half-year and quarterly financial statement publication as in the case of full year financial statement publication. On 17/2/2017, Ministry of Finance issued paper No. 01/VBHN/BTC that combined Circular 145/2016/ND-CP and Circular 108/2013/ND-CP, which provides guidance on punishment of administrative violations in fields of stock and stock market, effective from 15/12/2016. The maximum cash fine for entity in violation is 2 billions and 1 billion for individuals. These regulations have spurred the equality between listed and unlisted public firms, and aimed to protecting the lawful rights of shareholders, investors and improved the 20 transparency of stock market, deterred and dealt with violations on information publication in a timely manner. 4.1.2. The current status of information publication of listed firms in Vietnam With the sprawling stock market, the delay in information publication, especially financial statement publication, is on the rise. Figure 4.1. Violations on information publication from 01/2008 to 20/09/2017 Source: Author’s compilation In general, from the establishment event of Vietnamese stock market to 20/9/2017, most of regulations of State Stock Commission are published and are open to reviews before official issuance. The violations mostly center around reporting mode, information publication of public firms, listed firms; insiders‘ stock transaction, large shareholders, stock manipulation; and operations not consistent with license. State Stock Commission have issued punishment decisions for cases of delay in information publication. However, even with severe cash fines, violations in information publication are still hard to harness. Insiders at least have gained enormously by purchasing or selling stocks from delays in information publication on operation performances, and abnormal information. The delay in information publication, especially quarterly financial statements, has increasingly become common after 2010. Moreover, in the current context of abundance of bad news, firms tend to exaggerate their expenses to inflat the losses in the current periods so that they can reduce their expenses in the periods to come. This has t the ability to predict corporate income, profitability forecasting and so hinders the development of stock market. 21 4.1.3 Descriptive statistics on the information publication of listed firms in 01/2010- 06/2016 Figure 4.2. The number of firms publishing financial statements from 1/2010-6/2016 Source: Author‘s calculation The thesis also analyzes the time of publication for firms of different industries in Figure 4.4. Figure 4.4. Number of firms by industry that publish financial statements on time, late and early as compared to regulation (1/2010-6/2016) Source: Author‘s calculations Figure 4.5: Number of firms publishing financial statements from 1/2010-6/2016 compared to their own publication the previous year 22 Source: Author‘s calculations From figure 4.5 we can see that in Vietnam the time of financial statement publication is spread over 37 time periods, and unevenly scatters. However, it is clear that mainly firms issue on day 0. Figure 4.6: Number of firms publishing financial statements from 1/2010-6/2016 compared to their own publication in the previous year by industry Source: Author‘s calculations 4.2. Results regarding the time of financial statement publication and stock return (Model 1) 4.2.1. The relationship between good and bad news and the time of financial statement publication To quantify this relationship, the thesis adopts Nguyen (2010) by observing the changes of the parameters of market model across the windows in question [-90,90]. CAAR is 23 calculated using market model and event study and t-statistics. Tables 4.3 and 4.4 show that CAAR tends to be higher in short-term windows [-5,0] and long-term window [- 90,0] for both good and bad news prior to the publication of financial statements. This trend continues after the event but only short-term [0,5] and long-term windows [0,90] for good news. This could be because of the previous positive trend of the market. For bad news, the market responds negatively in a prolonged period, both for long and short- term windows. a. In short term Prior to the event (financial statement publication): in short term and with good news, t- statistics changes in days of financial publication, tend to be high for firms with good news and early publication. After the event: For good news after the publication t- statistics show no significance and not economically meaningful. T-statistics for good news is 0.55 for [0,1], 1.04 for [0,3] and 0.2 for [0,5]. Table 4.1. Relationship between good news, bad news and thet ime of financial statement publication in short term Good news ( UE>0) Bad news (UE<0) Days CAAR t-statistic Pr (T>t) Days CAAR t-statistic Pr (T>t) -5 0.002829 1.7657 (**) 0.0389 -5 0.007680 4.2254 (***) 0.0000 -4 0.002460 1.7006 (**) 0.0446 -4 0.006405 3.8582 (***) 0.0001 -3 0.000731 0.5534 0.2910 -3 0.002481 2.2718 (**) 0.018 -2 0.000536 0.5224 0.243 -2 0.003482 2.2919 (**) 0.011 -1 -0.000256 -0.2553 0.6007 -1 0.001876 1.590 (*) 0.0561 0 0.000093 1.0376 0.1498 0 -0.000117 -0.1471 0.5585 1 0.000675 0.5521 0.2905 1 -0.003866 -0.369 0.6442 2 0.001090 0.0760 0.4697 2 -0.005081 -0.4017 0.656 3 0.001682 1.0450 0.1481 3 -0.007497 -0.5354 0.7037 4 -0.000195 -0.1299 0.7774 4 -0.001292 -0.7538 0.5517 5 0.003319 0.2014 0.4202 5 -0.000518 -0.2766 0.6089 Notes: *** significant at 1% level; ** significant at 5% level; * significant at 10% level Source: Author’s calculations 24 b) In long term In long term with good news, t-statistics show statistical significance and tend to be high for good news being published early. Table 4.2. Relationship between good, bad news and the time of financial statement publication in long term Good news (UE>0) Bad news (UE<0) Days CAAR t- statistic Pr (T>t) Days CAAR t-statistic Pr (T>t) [-90,0] 0.00788 1.4176 (**) 0.0783 [-90,0] 0.03252 5.4547 (***) 0.000 [-60,0] 0.01811 3.8414 (***) 0.0001 [-60,0] 0.03579 6.9766 0.000 [-30,0] 0.01469 4.5379 (***) 0.0000 [-30,0] 0.02293 6.1678 (***) 0.000 [-10,0] 0.00627 3.1755 (***) 0.0008 [-10,0] 0.00898 3.9959 (***) 0.000 [10,0] 0.00021 0.1104 0.4561 [10,0] 0.00055 0.2468 0.4026 [30,0] 0.00309 0.5619 0.2871 [30,0] 0.00349 0.8158 0.2074 [60,0] 0.00298 0.5231 0.3005 [60,0] -0.00655 -1.0005 0.8714 [90,0] 0.00330 0.4978 0.3093 [90,0] -0.00930 -1.2910 0.9015 Notes: *** significant at 1% level; ** significant at 5% level; * significant at 10% level Source: Author’s calculations From tables 4.1 and 4.2, it is clear that for good news CAAR is highest on day -5 and + 5, and bad news CAAR is highest on day -5 and +3. If firms publish financial statements early, CAAR in [-5,0] is positive, regardless of bad news or good news, and this window also has high t-statistics and strong significance. On the other hand, the market responds negatively right after receiving the bad news (day 0). CAAR tends to be higher in short window [-5,0] and in long-term window [-90,0] for both good and bad news in the periods prior to financial statement publication. In summary, it can be concluded that the time of financial statement publication has impact on abnormal stock return, and managers strategize on information publication: bad news to be published late and good news to be published early. Therefore, there is evidence to accept H1. 25 4.2.2 Regressions on the link between time of financial statement publication and abnormal return The results presented in tables 4.3, 4.4 and 4.5 are as follows: For industry: This study employs Pooled OLS (from Bagnoli (2002)) to examine the link for each industry in cases of early, late and on time publication compared to last year publication. The findings show that the time firms publish financial statements late has impacted CAAR the most and statistically significant for 2 industries which are Consumer services and Consumer goods. There is no evidence that firms that publish earl have significant impact on CAAR. In case of on time publication, firms that deliver bad news are subject to punishment from market, especially Industrials, Consumer services. There is no evidence showing the impact of time of financial statement publication on CAAR for firms in industries like Energy & Gas and Technology. a. Regressions on the link between the time of financial statement publication and CAAR for firms that public late Table 4.3. Good news, bad news and CAAR of firms publishing information late CAAR ue late neg _ue late_ue negue_ue late _negue late _negue_ue _cons Sum Coef. 0.09724 -0.1816 -0.0944 0.0000 5.21E-06 -0.0139 -0.00001 0.28364 t 2.05 -4.29 -2.48 1.39 0.46 -0.23 -0.65 5.59 P>|t| 0.041 0.000 0.013 0.166 0.645 0.819 0.514 0.000 Obs 1081 R 2 0.1337 P > F 0.000 Basic Materials Coef. -0.0192 -0.4153 -0.0562 0.00002 -1.45E-05 0.14316 0.00001 0.41553 t -0.1 -2.83 -0.37 0.51 -0.23 0.64 0.13 2.1 P>|t| 0.918 0.005 0.71 0.613 0.822 0.524 0.897 0.037 Obs 131 R 2 0.1777 P > F 0.0225 Comsumer Goods Coef. 0.0939 -0.154 -1.231 -1.139 5.21E-06 -0.0409 7.4E-06 1.42407 t 1.98 -4.04 -1.93 -1.79 0.46 -0.7 0.38 2.23 P>|t| 0.048 0.000 0.053 0.074 0.645 0.482 0.704 0.026 Obs 153 R 2 0.1343 P > F 0.000 Consumer Services Coef. 0.10522 -0.5357 -0.4598 0.00001 2.01E-06 0.3512 -0.51560 0.63836 t 2.26 -9.77 -8.69 1.31 0.2 4.97 -9.59 10.31 P>|t| 0.024 0.000 0.000 0.19 0.841 0.000 0.000 0.000 Obs 72 26 R 2 0.1712 P > F 0.000 Health Care Coef. 0.08642 -0.2133 -0.1771 0.00005 6.17E-06 0.05118 -0.00004 0.34473 t 1.22 -3.21 -2.82 1.64 0.19 0.52 -0.86 4.57 P>|t| 0.223 0.001 0.005 0.100 0.852 0.605 0.392 0.000 Obs 32 R 2 0.1394 P > F 0.00 Industrials Coef. 0.19364 -0.3063 -0.1427 0.00012 -0.00011 0.05682 -0.00002 0.21200 t 1.26 -1.95 -0.93 1.11 -0.82 0.26 -0.14 1.24 P>|t| 0.209 0.054 0.354 0.269 0.412 0.796 0.888 0.217 Obs 595 R 2 0.1089 P > F 0.0402 Other Coef. 0.09547 -0.1743 -0.0993 0.00001 3.88E-06 -0.0095 -2.32E-06 0.28694 t 2.01 -4.28 -2.58 1.13 0.29 -0.16 -0.22 5.57 P>|t| 0.044 0.000 0.01 0.257 0.770 0.876 0.830 0.000 Obs 47 R 2 0.1336 P > F 0.000 Source: Author’s calculations b. Regression results on the link between the time of financial statement publication and CAAR for firms publishing early Table 4.4: Good news, bad news and the time of financial statement publication for firms publishing financial statements early CAAR ue early neg _ue early _ue negue _ue early _negue early _negue _ue _cons Sum Coef. 0.1141 0.00001 -0.1528 -6.1E-06 -0.1381 -0.07485 -5.2E-07 0.2588 t 2.39 0.89 -3.98 -0.49 -3.37 1.23 -0.02 5.15 P>|t| 0.017 0.374 0.00 0.627 0.001 0.218 0.982 0.000 Obs 571 R 2 0.1203 P > F 0.00 Basic Materials Coef. 0.2327 0.00004 -0.0040 -4.5E-06 -0.2162 -0.11244 -5.3E-05 0.0365 t 1.82 1.16 -0.04 -0.1 -1.84 -0.64 -0.61 0.26 P>|t| 0.07 0.247 0.97 0.924 0.067 0.52 0.543 0.798 Obs 91 R 2 0.1569 P > F 0.0315 Comsumer Goods Coef. 0.0871 0.00001 -0.1971 -1.1E-05 -0.1253 0.02524 -2.3E-06 0.3322 t 1.22 0.71 -3.29 -0.35 -1.91 0.25 -0.04 4.36 P>|t| 0.222 0.479 0.001 0.727 0.056 0.801 0.967 0.000 Obs 137 R 2 0.1294 27 P > F 0.00 Consumer Services Coef. 0.1142 0.00001 -0.1545 -6.2E-06 -1.39E-01 0.07576 0.02357 0.2474 t 2.39 1.14 -4.21 -0.59 -3.44 1.27 0.87 4.77 P>|t| 0.017 0.256 0.00 0.558 0.001 0.205 0.382 0.00 Obs 71 R 2 0.1206 P > F 0.00 Health Care Coef. 0.1135 0.00001 -1.2982 -6.3E-06 -1.36E-01 0.07340 -1.14682 1.4044 t 2.38 1.17 -2.02 -0.6 -3.37 1.23 -1.79 2.19 P>|t| 0.018 0.241 0.043 0.551 0.001 0.219 0.074 0.029 Obs 39 R 2 12.16 P > F 0.00 Industrials Coef. 0.1155 0.000004 -0.1348 -1.4E-06 -0.1303 0.05709 1.71E-05 0.2449 t 2.41 0.33 -3.43 -0.13 -3.19 0.93 1.29 4.77 P>|t| 0.016 0.742 0.001 0.898 0.001 0.352 0.196 0 Obs 571 R 2 12.10 P > F 0.00 Other Coef. -0.0394 0.00027 0.28563 -0.00022 -2.21E-01 -0.62653 -1,83E-04 0.5051 t -0.12 2.18 1.16 -1.16 -0.64 -1.42 -0.72 1.64 P>|t| 0.902 0.035 0.252 0.252 0.524 0.162 0.475 0.109 Obs 43 R 2 21.11 P > F 0.0505 Source: Author‘s calculation c. Regression results on the link between time of financial statements and CAAR for firms reporting on time compared to last year Table 4.5. The link between good news, bad news and the time of financial statement publication of firms publishing on time CAAR ue neg_ue negue_ue _cons Sum Coef. 0.11970 -0.13597 -0.10706 0.24073 t 2.53 -5.11 -3.91 4.99 P>|t| 0.011 0.00 0.00 0.00 obs 161 R-squared 0.1197 Prob > F 0.00 Comsumer Goods Coef. 0.11425 -0.11288 -0.05599 0.20357 t 2.41 -3.7 -1.42 4.27 P>|t| 0.016 0.00 0.157 0.00 R-squared 0.1142 obs 31 28 Prob > F 0.00 Prob > F 0.00 Industrials Coef. 0.21850 -0.21618 -0.00019 0.13713 t 1.43 -2.11 -2.23 0.87 P>|t| 0.154 0.036 0.028 0,385 obs 83 R-squared 0.164 Prob > F 0.0357 Source: Author‘s calculation 4.3. Research results (method 2) 4.3.1. Descriptive statistics Table 4.6. Descriptive statistics Variable Obs Mean Std. Dev. Min Max return 2723 0.23438 0.34309 -0.82 4.39 sue 2718 0.23713 0.26900 -3.59 2.81 size 2717 12.9190 1.07142 10.43 17.88 bm 2692 0.25705 0.25482 0 0.81 dvol 2726 21.2345 4.25703 9.98 28.31 ret12 2721 0.03535 0.27271 -0.81 14.23 da 2730 0.19194 0.15557 0 1 audit_s 2730 0.91978 0.27168 0 1 Source: Author‘s calculation 4.3.2. Analysis of correlation coefficients Table 4.7: Pearson correlation coefficient matrix Variable return sue size bm dvol ret12 da audit_s return 1.0000 sue 0.1971 1.0000 size -0.0990 -0.0853 1.0000 bm 0.0108 0.0216 -0.0447 1.0000 dvol 0.0157 0.0109 -0.0100 0.1023 1.0000 ret12 0.0103 0.0018 -0.0184 -0.0419 -0.0013 1.0000 da 0.0169 0.0271 0.0090 0.0125 -0.0177 -0.0039 1.0000 audit_s 0.0504 0.1031 -0.0154 -0.0247 0.0108 0.0051 0.0429 1.0000 Source: Author‘s calculation 29 The results from table 4.7 suggest that the absolute values of correlation coefficients between variables are smaller than 0.8, indicating that multicollinearity is not a major problem. 4.3.3. Multicollinearity test The author performs VIF (Variance Inflation Factor). The result is presented in table 4.8, showing that all coefficients are smaller than 10, which again confirms that multicollinearity should not be a major problem. Table 4.8: Results of multi-collinearity test Variable VIF 1/ VIF sue 1.02 0.981213 size 1.02 0.984628 audit_s 1.01 0.98698 dvol 1.01 0.988990 bm 1.01 0.990436 da 1.00 0.997305 ret12 1.00 0.997753 Mean VIF 1.01 Source: Author‘s calculation 4.3.4 Results of regressions Table 4.9: Results using Pooled OLS method Variable Coef. Std.Err. P>ǀtǀ sue 0.2699635*** 0.025555 0.000 size -0.0986713*** 0.027333 0.000 bm 0.0118245 0.022789 0.604 dvol -0.0014924 0.006468 0.818 ret12 -0.0010370 0.001508 0.492 da -0.0007827* 0.000443 0.077 audit_s 0.0345192 0.025411 0.174 _cons 0.1842679** 0.090302 0.041 Số quan sát 2178 F(7, 2170) 19.46 Prob>F 0.0000 30 R-squared 0.1152 Adj R-squared 0.1134 Notes: *** significant at 1% level; ** significant at 5% level; * significant at 10% level Source: Author’s calculations The results from table 4.9 show that sue, size and da are statistically significant. The study also employs FEM and REM to estimate panel data model. Table 4.10: Results using Fixed effects model Variable Coef. Std.Err. P>ǀtǀ sue 0.2404454*** 0.032484 0.000 size -0.110325*** 0.036533 0.003 bm 0.0134836 0.027614 0.625 dvol 0.0088709 0.009666 0.359 ret12 0.0007759 0.002614 0.767 da -0.000759 0.000487 0.119 audit_s 0.038842 0.036751 0.291 _cons 0.0177913 0.139474 0.899 Observation 2178 F test that all u_i = 0: F(481,1689) = 0.57 Prob > F R-squared : 0.7954 within 0.1296 between 0.2517 overall 0.1125 Notes: *** significant at 1% level; ** significant at 5% level; * significant at 10% level Source: Author’s calculations The results from table 4.10 show that FEM significant with Pvalue =0.0000, and the model explains 12.96% the variation in the dependent variable. Only sue and size are statistically significant. Table 4.11: Results using Random effects model Variable Coef. Std.Err. P>ǀzǀ sue 0.2699635*** 0.025555 0.000 size -0.098671*** 0.027333 0.000 bm 0.0118245 0.022789 0.604 dvol -0.001492 0.006468 0.818 31 ret12 -0.001037 0.001508 0.492 da -0.000783* 0.000443 0.077 audit_s 0.0345192 0.025411 0.174 _cons 0.1842679** 0.090302 0.041 Observation 2178 Wald chi2 (7) Prob > chi2 136.25 0.0000 R-squared : within 0.1284 between 0.2646 overall 0.1591 Notes: *** significant at 1% level; ** significant at 5% level; * significant at 10% level Source: Author’s calculations The results from table 4.11 show that REM model is significant with pvalue =0.0000, and the explanatories explain 12.84% of the variation in the dependent variable (stock return). In this model, sue, size and da are statistically significant. The study continues with Breusch & Pagan, Hausman tests to decide which model to choose (among Pooled OLS, FEM, REM). 4.3.5. Tests for model selection Table 4.12: Summary of tests for model selection Test Pooled-OLS & FEM Pooled-OLS & REM FEM & REM F - test F (481, 1689) = 0.57 và Prob > F= 0.7954 Breusch – Pagan test Chibar2 (01) = 0.00 Prob > chibar2 = 0.0447 Hausman test Prob>chi2 = 0.6520 Conclude Choose Pooled-OLS Choose REM Choose REM Source: author‘s calculations b/Test for heteroskedasticity: the result shows that there is heteroskedasticity c/Test for autocorrelation: the result shows that theire is no autocorrelation 32 d/Address heteroskedasticity: the study continues with REM method with robust option to address heteroskedasticty Table 4.13: Regression results after addressing heteroskedasticity Variable Coef. Std.Err. P>ǀtǀ sue 0.2699635*** 0.041146 0.000 size -0.098671*** 0.027148 0.000 bm 0.0118245*** 0.001729 0.000 dvol -0.001492 0.006909 0.829 ret12 -0.001037 0.001416 0.464 da -0.000783* 0.000448 0.081 audit_s 0.0345192 0.024845 0.165 _cons 0.1842679* 0.095204 0.053 Obs 2178 Wall chi2(7) 327.11 Prob > chi2 R-squared : 0.0000 within 0.1284 between 0.2646 overall 0.1594 Notes: *** significant at 1% level; ** significant at 5% level; * significant at 10% level Source: Author’s calculations Table 4.13 shows that among the 7 regressors, 3 (sue, size, bm) are significant at 1% level and 1 (da) at 10%. Dvol, ret12 and audit_s are those that are not statistically significant. Bm and sue are positively and correlated with stock returns, and size and adjusted accounting income have negative correlation with stock returns. The other variables are not significant. 4.3.6 Results and discussion 4.3.6.1. Statistically significant variables SUE: this variable is positively correlated with stock return and significant at 1% level. The coefficient is 0.26996, meaning that as standardized income increases by 1 unit, stock return increases by 0.26996 unit, ceteris paribus. Hypothesis H8 about the positive 33 link between stock return and standardized unexpected earnings is accepted. This result is also in line with studies such as Chordia (2006), Vinh and Phuong (2014). Size: this variable is negatively correlated with stock return and significant at 1% level. The coefficient is -0.098671, meaning that when size increases by 1 unit, stock return decreases by 0.098671 unit. This is also consistent with Chordia (2006, 2013) and Vinh and Phuong (2014). The hypothesis H2 is accepted. BM: bm means the ratio of book to market value. The result shows that there is a positive relation between book to market ratio and significant at 1% level. The coefficient is 0.011824, meaning that when book to market ratio increases by 1 unit, stock return increases by 0.011824 unit, ceteris paribus. Therefore, the hypothesis H3 is accepted, which is consistent with Yuenan Wang and Amalia Di Iorio (2007), Vinh and Phuong (2014), Nguyen Anh Phong (2015). DA: managers should abide by the laws and at the same time be flexible in preparing financial statements that reflect firm‘s financial health the best. They should not illegally perform earnings management as in Friedlan (1994). The regression result shows that there is a negative relationship between discretionary accruals and stock returns at 10% level. The coefficient is -0.000783, meaning that when discretionary accruals increase by 1 unit, stock return decreases by 0.000783 unit. This is consistent with hypothesis H6. This is also compatible with the realit in emerging stock markets like Vietnam, with most investors searching for profits from firms as soon as possible, especially right in the year they hold the stock rather than years later in the future. This result is consistent with the findings from Erlynda Y Kasim (2013) for Indonesian stock market. 4.2.6.2. Statistically insignificant variables DVOL: Transaction volume in money is the volume of securities sold during the day, calculated as the unadjusted price at closing multiplied by the number of shares traded in the day, reflecting trading activities of investors in a certain time period. As the number of shares traded per day can be very large, the transaction value in money used in the research is also large. Therefore, this variable is calculated as its natural logarithm (Bernnan, 1998). The analysis of this factor is to confirm the strength of the current trend, or to determine the potential reversal of stock prices. The results show that there is a negative relationship between the transaction volume in money and stock return but with no statistical significance (P.value = 0.829). Therefore, the research results have not 34 proved the impact of traded volume in money and stock return. There is insufficient statistical evidence to accept the H5 hypothesis, and this result is consistent with that of Vinh and Phuong (2014). RET12: The RET12 variable is stock return momentum. Based on this index, investors can judge the trend of stock returns in the present and future. The research results have not proved that there is a positive relationship between the momentum of stock return and stock return (P.value = 0.464). Therefore, there is insufficient statistical evidence to accept the hypothesis H4. The result is consistent with the results of Brennan (1998), Chordia (2006) and in contrast with the results of Vinh and Phuong (2014). Audit-s: This variable represents the size of the auditing firm (whether auditing firm belongs to Big6). Regression results show that its coefficient positive and not statistically significant (P.value = 0.165). This result shows that investors have not paid much attention to whether the listed companies are audited by the Big 6 companies or not. In short, there is insufficient statistical evidence to accept the H7 hypothesis which claims that the influence of the size of the audit firm has impact on stock return. The results in section 4.1 (descriptive statistics) show that the ratios of financial statements having exception audit clause are pretty similar between the companies audited by Big6 (42.77%) and not Big6 (57.23%). This shows that investors have not paid attention to whether the financial statements of companies are audited by one of the largest 6 auditing companies in Vietnam. This result was also found in the Indonesian stock market as in Erlynda Y. Kasim (2013). Table 4.14. Summary of research results Variable Realistic expectations Result Size - (***) consistent with the hypothesis, and in consistent with Chordia (2006, 2013), and the same sign as in Vinh and Phuong (2014) BM + (***) consistent with the hypothesis, and in consistent with Yuenan Wang and Amalia Di Iorio (2007), Vinh and Phuong (2014), Nguyen Anh Phong (2015) RET12 - consistent with the hypothesis, in consistent with Vinh and Phuong (2014), but in contrast with Brennan (1998), Chordia (2006). RET12 is insignificant. 35 Dvol - Inconsistent with the hypothesis, and this result is in contrast with Vinh and Phuong (2014). This result shows that DVOL is insignificant. DA _ (*) Consistent with the hypothesis and is significant, and consistent with Erlynda Y. Kasim (2013). Audit-s + Consistent with the hypothesis and is significant, and consistent with Erlynda Y. Kasim (2013), showing that audit_s is insignificant. Sue + (***) Consistent with the hypothesis and is significant, and consistent with Chordia (2006), Vinh and Phuong (2014). Notes: *** significant at 1% level; ** significant at 5% level; * significant at 10% level Source: Author’s summary CONCLUSION OF CHAPTER 4 This chapter presents and discusses research findings. CHAPTER 5: CONCLUSION AND POLICY IMPLICATIONS 5.1. Conclusion The thesis draws following conclusions: 5.1.1. Research objectives and hypotheses The thesis has finalized the testing of the following hypotheses: Table 5.1: Synthesis of tested hypotheses Hypothesis Content Reject/Accept H1 There is a link between time of financial statement publication and CAAR. Managers strategize on information publication, meaning that bad news to be published late and good news to be published early. Accept H2 There is a negative link between firm size and stock return when financial statements are published Accept H3 There is a positive link between book to market ratio and stock return when financial statements are published Accept H4 There is a positive link between return momentum and stock momentum when financial statements are published Does not accept H5 There is a positive link between transaction volume measured in monetary unit and stock return when financial statements are published Does not accept 36 H6 There is a negative link between discretionary accruals and stock returns when financial statements are published Accept H7 There is a positive link between auditor‘s size and stock return when financial statements are published Does not accept H8 There is a positive link between standardized unexpected earnings and stock returns when financial statements are published Accept Source: Author’s summary 5.1.2. Research results Domestic and foreign studies have shown that information on financial reports is considered by managers, investors and regulators as one of the sources of official and important information. Based on the study of the relationship between publication time and financial factors, the dissertation has dealt with the following issues: Firstly, the thesis has provided the theoretical foundations of efficient market, agency theory, behavioral finance theory, information concepts, financial information and publication time. In addition, the dissertation collects evidence from domestic and international research on the impact of time and information on the published financial statements. Secondly, the thesis has synthetic the method of event research (with regard to the time of financial statement disclosure), to study the stock return of each listed firm with late and early publication, for the whole sample and each individual industry. + In addition, the thesis has proposed the research hypotheses, collect data to perform regression analysis as well as test relevant hypothesess. + The dissertation has systematized the theoretical basis for quantitative research methods to be conducted for this study. Thirdly, the dissertation has conducted research on the impact of timing of financial statements disclosure and the financial factors on stock returns. Research has addressed the following issues: (a) The timing of financial statement disclosure has an impact on abnormal stock returns of listed firms and executives of listed firms do implement disclosure strategy, ie, good information from the financial statements is released early and bad news is released late. + By using the event method and the adjusted market return model according to Brown and Warner (1985), Hedge and McDermott (2003), Denis et al. (2003), Gregoriou and 37 Ioannidis (2006), Nguyen (2010) during 181 business days around the date of publication of income, the following results are obtained: For good news CAAR is highest on day -5 and + 5, and bad news CAAR is highest on day -5. If firms publish financial statements early, CAAR in [-5,0] is positive, regardless of bad news or good news, and this window also has high t-statistics and strong significance. On the other hand, the market responds negatively right after receiving the bad news (day 0). CAAR tends to be higher in short window [-5,0] and in long-term window [-90,0] for both good and bad news in the periods prior to financial statement publication. For industry: This study employs Pooled OLS (from Bagnoli (2002)) to examine the link for each industry in cases of early, late and on time publication compared to last year publication. The findings show that the time firms publish financial statements late has impacted CAAR the most and statistically significant for 2 industries which are Consumer services and Consumer goods. There is no evidence that firms that publish early have significant impact on CAAR. In case of on time publication, firms that deliver bad news are subject to punishment from market, especially Industrials, Consumer services. There is no evidence showing the impact of time of financial statement publication on CAAR for firms in industries like Energy & Gas and Technology. (b) The dissertation, using three regression models Pooled OLS, FEM, REM, has identified and quantified the impact of information disclosed on the financial reports on stock returns in Vietnam. Specifically, factors such as company size, book to market value , standardized unexpected earnings, discretionary accruals significantly impact stock returns. In addition, the study cannot find statistical evidence about the relationship between stock return and transaction volume in cash, the momentum of stock return and the size of the audit firm. Fourthly, the thesis has made some recommendations related to the timing of publication of financial statements and financial information on financial statements, thereby enhancing the stock returns of listed firms. + Based on the results of regression and discussion of research results, the thesis proposes 3 groups of recommendations for the following stakeholders: 38 * Investors: When it comes to the time of financial statement disclosure of listed firms in short and long term, investors should pay attention to firms in industries of Consumer Services and Consumer Goods when late financial statements are released. The information disclosed on the financial statements, such as book to market value, company size, abnormal income fluctuations, especially discretionary accruals, etc. * Listed companies: When it comes to the time of financial statement disclosure, the disclosure should be in accordance with the State regulations on disclosure, ensure good grasp and strict management of internal information, timely auditing of financial statements, create prestige and trust for investors as well as the authorities. It is not always easy to expand the company to garner high stock returns, and instead firms need to effectively manage the existing resources in the current economic situation, focus on public relations (IR), corporate social responsibility, ... * The authorities: need to have separate regulation on information disclosure for listed firms with high market value, and impose strict delisting on firms that turn in financial statements late... 5.2 Recommendations for future research The thesis points out the limitations of the study and suggests directions for future research, research sample and other factors affecting the stock market. CONCLUSION CHAPTER 5 In Chapter 5, the thesis concludes and proposes recommendations to investors, managers of listed firms and state authorities to enhance stock returns, step by step improve the transparency of stock market, contribute to the effective operation of stock market. The thesis also outlines the new research contributions and presents the limitations of the research and further research. 39 LIST OF AUTHOR’S PUBLICATION (1) Nguyen Thi Ngoc Diep (2017): ―The timing of financial statement publication and stock return: the case of Vietnam‖. International Conference for Young Researchers in Economics and Business (ICYREB 2017). ISBN: 978-604-84-2640-8 (2) Nguyen Thi Ngoc Diep (2017): ―Financial reporting information and impact on stock returns‖. Financial Journal Vol 2-2017 (657) .pp 39 (3) Nguyen Thi Ngoc Diep, Bui Van Thuy, Hoang Thi Quynh Anh (2017): ―The Effect Voluntary Disclosure, Macro and Micro Factors on Stock Returns: Evidence from Vietnam‖. Journal of Science of Lac Hong University 2017- ISSN : 2525-2186. Registration authentication. (4) Nguyen Thi Ngoc Diep, Nguyen Thanh Lam (2017): ―The Impact of Earnings Information on Financial Statements on Stock Returns in Vietnam‖. Journal of Science of Lac Hong University 2017- ISSN : 2525-2186. registration authentication (5) Pham Ngọc Toàn, Nguyen Thi Hang Nga, Nguyen Kim Nam, Nguyen Thi Ngoc Diep (2016): ―Effects of audit firm size and auditor charactereristics on firms‘ discretionaty accrual management‖. Proceedings of international conference on accounting & finance 2016 (ICOAF 2016). ISBN 978-604-84-1563-1 (6) Nguyen Thi Ngoc Diep, Nguyen Thanh Lam (2017): ―Determinants of Liquidity of Commercial Banks in Vietnam in the Period 2009-2016‖. International Journal of Scientific Study, ISSN: 2321-6379, Vol 5. DOI: 10.17354/ijssSept/2017/45. (7) Nguyen Thanh Liem, Nguyễn Thị Canh, Nguyen Thi Ngoc Diep (2017): ―Corporate debt maturity structure: quantile regression and Oaxaca-Blinder decomposition approaches‖. International Conference for Young Researchers in Economics and Business (ICYREB 2017). ISBN: 978-604-84-2640-8 (8) Nguyen Thi Ngoc Diep, Nguyen Kim Nam, Nguyen Thi Hang Nga (2016): ―The impact of foreign direct investment and financial development on economic growth: Evidence Asean countries‖. Proceedings of the Foreign Trade University. ISBN: 987-604-73-9348-8 (9) Nguyen Thi Ngoc Diep, Nguyen Minh Kieu (2015): ― Effects of Specific Banking Fctors on Credit Rick of Vietnam‘s Commercial Bank‖. Journal of Economic Development. Vol 3 - ISSN 1859-1124. (10) Nam K.Nguyen, Nga T.H.Nguyen, Diep T.N.Nguyen (2015) : ―The Impact Of Business And Consumer Confidence On Stock Market Risk Premiums: Evidence From Vietnam‖. Asian Journal of Management Sciences. ISSN 2348-0351. (11) Nguyen Thi Ngoc Diep, Nguyen Minh Kieu (2013):― Financial factors affecting investment decision by individual investors at HCMC trading floor‖. Science of Open Unversity. Vol 3 – ISSN 1859-3453 THÔNG TIN TÓM TẮT NHỮNG ĐÓNG GÓP MỚI VỀ MẶT HỌC THUẬT VÀ LÝ LUẬN CỦA LUẬN ÁN Tên đề tài: THÔNG TIN TÀI CHÍNH TÁC ĐỘNG ĐẾN SUẤT SINH LỜI CHỨNG KHOÁN CỦA CÁC CÔNG TY NIÊM YẾT TẠI THỊ TRƢỜNG CHỨNG KHOÁN VIỆT NAM Chuyên ngành: Tài chính - Ngân Hàng. Mã số: 62.34.02.01 Cơ sở đào tạo: Trƣờng Đại Học Ngân Hàng TP. Hồ Chí Minh Người hướng dẫn khoa học: PGS. TS. LÝ HOÀNG ÁNH Luận án này thực hiện để nghiên cứu thông tin tài chính ảnh hưởng đến suất sinh lời cổ phiếu khi công bố BCTC. Với những kết quả đạt được từ mô hình thực nghiệm, về mặt học thuật, luận án mang mang lại đóng góp mới trên cơ sở bổ sung vào khoảng trống nghiên cứu thông qua cung cấp bằng chứng thực nghiệm về tác động của thời gian công bố BCTC và các thông tin tài chính tác động đến suất sinh lời. (4) Nghiên cứu được thực hiện trên phạm vi toàn TTCK VN và có ưu điểm là nghiên cứu trên từng nhóm ngành cụ thể, dựa trên phương pháp nghiên cứu sự kiện, lấp đầy các khoảng trống nghiên cứu về phạm vi này. Kết quả cho thấy CAAR có xu hướng cao hơn trong cả khoảng thời gian ngắn hạn [-5;0] và trong khoảng thời gian dài hạn [-90;0] cho cả tin tốt và tin xấu trong giai đoạn trước sự kiện công bố BCTC. Đối với các ngành: Thời gian các CTNY công bố BCTC trễ có tác động đến CAAR trong hầu hết các ngành, phản ứng tiêu cực mạnh nhất và có ý nghĩa thống kê đối với hai ngành là Dịch vụ tiêu dùng và Hàng hóa tiêu dùng; Đối với trường hợp công bố BCTC sớm, chưa có bằng chứng thực nghiệm về các công ty có thời gian công bố sớm tác động có ý nghĩa đến CAAR. Nếu các công ty có thời gian công bố BCTC sớm mà các thông tin chứa đựng trong đó là xấu thì cũng nhận được phản ứng tiêu cực của thị trường và đều có ý nghĩa thống kê, mạnh nhất là đối với ngành Chăm sóc sức khỏe, kế đến là ngành Hàng hóa tiêu dùng, Dịch vụ tiêu dùng; Trong trường hợp công bố đúng thời gian thì thông tin về thời gian công bố BCTC có tác động dương đến CAAR và có ý nghĩa thống kê, nhưng những thông tin trên BCTC là xấu thì cũng chịu sự trừng phạt của thị trường, mạnh nhất ở nhành Công nghiệp, sau đó là ngành Hàng hóa tiêu dùng. Chưa có bằng chứng thống kê cho thấy tác động của thời gian công bố BCTC đến CAAR đối với ngành năng lượng & khí đốt và ngành công nghệ. (5) Từ kết quả nghiên cứu luận án đã xác định và định lượng được các thông tin công bố trên BCTC có tác động đến SSL chứng khoán của các CTNY trên TTCK Việt Nam, đặc biệt là lợi nhuận kế toán điều chỉnh có tác động ngược chiều và có ý nghĩa thống kê đến suất sinh lợi chứng khoán. (6) Cuối cùng, luận án cũng đưa ra những hạn chế và một số khuyến nghị liên quan đến thời gian công bố BCTC và các TTTC trên BCTC, qua đó nâng cao SSL đối với nhà đầu tư, đối với các CTNY và đối với cơ quan quản lý nhà nước. Người hướng dẫn khoa học Nghiên cứu sinh PGS. TS. LÝ HOÀNG ÁNH NGUYỄN THỊ NGỌC DIỆP A SUMMARY OF INFORMATION ON NEW CONTRIBUTIONS OF THE THESIS TILE OF THE THESIS: THE IMPACT OF FINANCIAL INFORMATION ON STOCK RETURN OF VIETNAMESE LISTED FIRMS Major : Finance and Banking Code: 62.34.02.01 Institution : Banking University Hochiminh City Science instructor: Assoc. Prof. Dr. Ly Hoang Anh This dissertation is to study the impact of financial information on stock returns when financial statements are announced. Academically, the thesis brings new contributions on the basis of addition to the research gap by providing empirical evidence on the effect of timing of financial statement announcement and financial information on firms‘ profitability. (1) The research was conducted using a sample of all Vietnamese listed firms and event study and it is advantageous to study specific industries. The results sugget that CAAR tends to be higher in short window [-5;0] and in long window [-90;0] for both good news and bad news in the period prior to the announcement. With related to industries: the time when firms announce financial statements late tend to have negative impact on CAAR for most industries, especially for two industries namely Consumer Services and Consumer Goods; For firms that pubish financial statements early, no empirical evidence is recorded that early announcement of financial statements has impact on CAAR. If firms publish information early and information contained is bad news then firms tend to receive statistically significant negative feedback from market, especially strongest is Industrials, then Consumer Goods. There is no evidence showing the impact of the timing of financial statement on CAAR in Energy and Gas and Technology industries. 2) The empirical findings have identified and quantified that the information contained in financial statement has impact on stock return of listed firms in Vietnam, especially accounting profit (xem lại) has a significant and negative impact on stock return. 3) Finally, the thesis outlines several limitations and some suggestions regarding the timing of financial statements and the content of financial statements, thereby improving stock returns for investors, for both listed firms and the authorities. Science instructor Research student Assoc. Prof. Dr. Ly Hoang Anh Nguyen Thi Ngoc Diep