The Different Dividend Signaling Effect under Tax Deduction around Ex-Right Day: Evidence from Taiwan Stock Exchange
Jui-Cheng Hung
Round 1
Reviewer 1 Report
Comments:
1. The authors employ market model to estimate the expected return of individual stocks and then compute the abnormal returns for Taiwan listed companies during the research period of 2014 to 2016. In this sense, the estimation of abnormal returns is based on the assumption of CAPM with one risk factor. As the Fama and French three factor model is widely used in empirical studies, are the results consistent when using Fama and French three factor model to compute abnormal returns?
2. The research period consists of the daily returns from 2014 to 2016. However, the significant depreciation of RMB caused a sharp drop in Taiwan's stock market during 2015 August. Do the empirical results in this article need to consider the impact of this event?
Author Response
The author is very appreciated for the suggestion of the review, and the reply is as follows:
1.Please refer to the manuscript in line 322-337.
The abnormal return in the event study method is defined as the actual rate of return minus the expected rate of return during the event analysis period. There are several methods used to measure the expected return of stocks in the research. Some use the stock's own historical data as the normal rate of return; there are normal return adjusted by the market's single-factor return (MACKINLAY, 1997), such as: constant-mean- return, mean-adjusted returns, market-adjusted returns, etc.; there are also models that consider multiple factors, such as the APT proposed by Born (1984) and the three factor mode proposed by Fama and French (1993).
Because biased expected returns can lead to erroneous results in event studies. However, the results of Brown and Warner (1980), Brown and Weinstein (1985), Chandra, Moriarty and Willinger (1990), Binder (1998) all found that because the multifactor model is limited by the estimation procedure and the choice of factors must be very Correctness makes the expected return estimates more correct than the one-factor market model. Otherwise event studies using multifactor models are no more powerful than those using market models. Therefore, this study adopts the market model to calculate the expected rate of return of the stock.
2.Please refer to the manuscript in the 3rd part, in line 284-337.
MACKINLAY (1997) pointed out that if the event research method adopts a large sample size, the problem of statistical bias will be reduced. Therefore, the event research method design of this study is based on the steps of the literature, with a period of 3 years and a large sample of 1902 samples. In addition, the 1902 pieces of data have the same ex-rights event. In this way, the impact on the stock price caused by this event has been evenly offset, and the rest is the focus of this study - the correlation of dividends. Tax Policy and Fee Policy. Therefore, for example, the impact of RMB depreciation on Taiwan stocks in 2015 is not the focus of this study, and the extent of the impact of the empirical conclusion does not reach a statistically significant level.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Major comments
1. Need to take a higher level view in the abstract. What are you studying and why is it important? Why conduct the study in Taiwan? What method is used and what are your results? Then why is it important?
2. What is your research question? This study seems to lack a clear research question, making it difficult for the reader to assess what we learned from the study
3. What is your hypothesis? Your hypothesis is unclear as is your research question.
4. Consider Binder (1998) and MacKinlay (1997) regarding event study methodology
5. Should consider the Efficient Market Hypothesis, that all available information is in the stock price. Yes, it would then make sense that the stock price would change with the event of a dividend tax change, which would make dividends more valuable.
6. End the literature review with your contribution
7. Literature is fairly old, with most of the papers being from the 1990s. Please add five or so papers from the last five years.
8. Wy 2014, 2015, and 2016?
9. You should use a smaller event window as well, perhaps (-2,2). Use this event window in conjunction a larger one as the two models replacing the models you have.
10. Use the Fama-French 4 factory model, adding company size, value versus growth, and momentum
11. Explain the relevance of the before versus after the event date. The values for the after event date are consistently higher. This finding should be highlighted.
12. Line 393 take out correlated and put in positively impacts.
13. Explain the implication behind the positive coefficients for REDUCT and PR.
14. R-squared is very low suggesting omitted variable bias
Minor comments
1. In abstract, change deduction to reduction
2. Rewrite first sentence “Dividend income taxes are… take out the hyphen in the first sentence, take out both
3. Take out intense in the second sentence
4. Line 35, who started to implement?
5. Line 83 is a sentence fragment
6. Line 84, who is it in your statement, it was found.
7. Line 87, who should amend the income tax laws?
8. Line 114, I don’t know what this means, “The spirit is to wish to plan towards the individual gross income system.”
9. Sentence on line 115 needs to be rewritten
10. Line 118 -123 need to be rewritten
11. Line 135 is a sentence fragment
12. Line 172 is a sentence fragment
13. Line 178, their not the conclusion
14. Line 189 take out in their research paper
15. Line 196, take out the paper notes and replace with the results indicated
16. Line 216, briefly explain in one line the abstention hypothesis and line 225 equilibrium theory of trading volume
17. Line 229 does not make sense. “the level of the company's declared dividend is related to the level of the company's market risk” Instead, the company’s declared dividend is related to their dividend policy and cash flow
18. Line 302, what is a strain variable
19. Line 336 what does PR stand for.
20. Put the stars in table 2 next to the coefficient not the standard error
21. Line 416, change the sentence to say, the impact of these reduced tax cuts on abnormal returns is the focus on the study
Author Response
The authors are very appreciated for the suggestion of the review, and the reply are as following:
1.The author has made revisions to this article. Please refer to the manuscript in line 10-24.
Dividend tax policy is one of the important tools of government taxation. Observing the dividend tax policy and the behavior of stock prices around ex-rights will not only shed light on investment strategies, but also give us a clearer understanding of the microstructure of the capital market. Taiwan went through dividend tax policy and National Health Insurance (NHI) supplementary premium changes from 2014 to 2016. Therefore, this paper adopts the event study method to conduct empirical research on this major event period. The research conclusion points out: (1) During the research period, the company has a positive cumulative abnormal return before and on the ex-right day; and there is a negative cumulative abnormal return after the ex-right day. (2) When the tax reduction effect is more favorable to investors, there will be only a positive relationship with positive abnormal returns. (3) There is no statistical significance between the dividend tax reform policy and the negative abnormal return after ex-rights. The empirical results of this paper can help to better understand the pricing process of stocks by market microstructure systems such as dividend tax policies, and help build a more stable stock market transaction structure. On the other hand, investors and companies can also gain their own investment or dividend policy inspiration from this research.
2.The author has made revisions to this article. Please refer to the manuscript in line 31-38.
The rapid evolution of globalized economic policies has led to the continuous reform of tax systems by governments in various countries in response to changes in the environment. The tax policy of dividend income not only affects tax efficiency, social equity, and national fiscal revenue; on the other hand, it is also a research topic that the microstructure theory of finance field attaches great importance to. Since dividend income tax is an important change in investment transaction rules, understanding how this system change affects the ex-rights process and outcome of securities assets will help clarify the behavior patterns of stock market participants.
3.The author has made revisions to this article. Please refer to the manuscript in line 185-187,271 and 280.
H1a:The stock has a positive abnormal return on the ex-rights date
H1b: The stock has a positive cumulative abnormal return before the ex-rights date
H1c: The stock has a negative cumulative abnormal return after the ex-rights date
H2：When the tax deduction effect is more favorable to investors, there will be obvious positive abnormal returns.
H3：The 2-generation NHI premium has a negative correlation with the positive abnormal return on the ex-rights date.
4.The author has made revisions to this article. Please refer to the manuscript in line 322-337. and the 3rd part, in line 284-337.
The abnormal return in the event study method is defined as the actual rate of return minus the expected rate of return during the event analysis period. There are several methods used to measure the expected return of stocks in the research. Some use the stock's own historical data as the normal rate of return; there are normal return adjusted by the market's single-factor return (MACKINLAY, 1997), such as: constant-mean- return, mean-adjusted returns, market-adjusted returns, etc.; there are also models that consider multiple factors, such as the APT proposed by Born (1984) and the three factor mode proposed by Fama and French (1993).
Because biased expected returns can lead to erroneous results in event studies. However, the results of Brown and Warner (1980), Brown and Weinstein (1985), Chandra, Moriarty and Willinger (1990), Binder (1998) all found that because the multifactor model is limited by the estimation procedure and the choice of factors must be very Correctness makes the expected return estimates more correct than the one-factor market model. Otherwise event studies using multifactor models are no more powerful than those using market models. Therefore, this study adopts the market model to calculate the expected rate of return of the stock.
5.The author has made revisions to this article. Please refer to the manuscript in line 87, 299-302, 303-306,87 and 280.
6.The author has made revisions to this article. Please refer to the manuscript in 2nd literature review part,and in line 151-281.
7.The author has made revisions to this article. Please refer to the manuscript in line 564-617.
8.The author has made revisions to this article. Please refer to the manuscript in line 71-83.
Based on the above theoretical and practical background, this study intends to use the ex-right day of listed companies in 2014, 2015 and 2016 in Taiwan exchange market as the event day, and uses the event study method to explore whether the market anomaly of average excess returns will still occur in listed companies around the ex-rights day after two tax policy reforms in Taiwan.
Therefore, the dividend deductible personal comprehensive income tax rate is the most favorable at 100% in 2014, and 50% in 2015 and 2016. Dividends are subject to a 2-generation NHI premium ratio of 2% in 2014 and 2015 and 1.91% in 2016. Based on the balance of the sample size consideration, this three-year period has fully captured the 2 events that the purpose of this study is to demonstrate - tax reform and additional NHI premiums. If the period is too long, the number of sample size of each event will be unbalanced, which may lead to biased conclusions.
9.The author has made revisions to this article. Please refer to the manuscript in line 303-312.
Usually the event window will be longer than the date of the event, including a period of time before and after the event. Because the information content in the period before and after the event is relatively sufficient, investors can better capture the signs before the event and the impact after the event. In particular, Taiwan's capital market is not a strong efficiency market, and the stock market usually cannot fully respond to specific events in a very short period of time.
The design of the event window of this study refers to the research method of Boehmer, Masumeci & Poulsen (1991), and after pre-checking the data, the abnormal return of the stock price of the observed sample starts about 10 days before the ex-rights date and ends 10 days after the ex-rights date. Therefore, this study The designed event window is in the range of (-10,10).
10.The author has made revisions to this article. Please refer to the manuscript in line 322-337.
The abnormal return in the event study method is defined as the actual rate of return minus the expected rate of return during the event analysis period. There are several methods used to measure the expected return of stocks in the research. Some use the stock's own historical data as the normal rate of return; there are normal return adjusted by the market's single-factor return (MACKINLAY, 1997), such as: constant-mean- return, mean-adjusted returns, market-adjusted returns, etc.; there are also models that consider multiple factors, such as the APT proposed by Born (1984) and the three factor mode proposed by Fama and French (1993).
Because biased expected returns can lead to erroneous results in event studies. However, the results of Brown and Warner (1980), Brown and Weinstein (1985), Chandra, Moriarty and Willinger (1990), Binder (1998) all found that because the multifactor model is limited by the estimation procedure and the choice of factors must be very Correctness makes the expected return estimates more correct than the one-factor market model. Otherwise event studies using multifactor models are no more powerful than those using market models. Therefore, this study adopts the market model to calculate the expected rate of return of the stock.
11.The author has made revisions to this article. Please refer to the manuscript in line 460-464.
The values for the after event date are consistently higher than before the event date. It shows that investors tend to sell their shares after receiving dividends ex-rights. This phenomenon is consistent with Eyup Kadioglu & Ayhan Kir-bas (2021), Grinblatt, Masulis and Titman (1984). We also verified the research hypotheses H1b and H1c.
12.The author has made revisions to this article. Please refer to the manuscript in line 484-491.
Therefore, it can be considered that the dividends received by investors for participating in ex-rights will be tax refunded when the individual is paid. This result supports the hypothesis H2 proposed in this paper. When the tax deduction effect is more favorable to investors, there will be obvious abnormal returns.
The dependent variable of Model II is the abnormal rate of return on the ex-right day. The estimated value of β2 (HEALTH)in mode 2 is -2.725, the statistical test value of the t value is -1.789, and the p value reaches the 10% test level, This result also supports the hypothesis H3 proposed in this paper.
13.The author has made revisions to this article. Please refer to the manuscript in line 405, 414, 533-538.
14.The author has made revisions to this article. Please refer to the manuscript in line 495-502, 552-554.
As for the phenomenon that the R-squared of the three-model is lower, it is often found in the studies on abnormal returns by the event study method. (Junnosuke, Shino, Takahashic (2022), Kreidl (2020), Lin, Chen & Lee (2001)). The possible reason is that the value of the abnormal return has a small change, while the independent variable has a large change, thus causing the X-axis of the regression to be nearly parallel. In addition, adding independent variables is also a way to improve R 2. However, the F values of the three regressions in this study were all statistically significant, so the three regression models were well suited.
In addition, the three regression models in this study have the phenomenon that R-squared is very low, which can be improved by adjusting the research model.
Minor Commnets:
1.The author has made revisions to this article. Please refer to the manuscript in line 18.
2.The author has made revisions to this article. Please refer to the manuscript in line 39-42.
3.The author has made revisions to this article. Please refer to the manuscript in line 43-45.
4.The author has made revisions to this article. Please refer to the manuscript in line 46.
5.The author has made revisions to this article. Please refer to the manuscript in line 87.
6.The author has made revisions to this article. Please refer to the manuscript in line 10-106.
7.The author has made revisions to this article. Please refer to the manuscript in line 109.
8.The author has made revisions to this article. Please refer to the manuscript in line 136-142.
9.The author has made revisions to this article. Please refer to the manuscript in line 155-156.
10.The author has made revisions to this article. Please refer to the manuscript in line 234.
11.The author has made revisions to this article. Please refer to the manuscript in line 236.
12.The author has made revisions to this article.
13.The author has made revisions to this article.
14.The author has made revisions to this article.
15.The author has made revisions to this article.
16.The author has made revisions to this article. Please refer to the manuscript in line 242-247.
17.The author has made revisions to this article. Please refer to the manuscript in line 251-254.
18.The author has made revisions to this article. Please refer to the manuscript in line 378.
19.The author has made revisions to this article. Please refer to the manuscript in line 414-420.
20.The author has made revisions to this article. Please refer to the manuscript in line 504.
21.The author has made revisions to this article. Please refer to the manuscript in line 517-520.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Need to make the following changes in the hypotheses,
For H1 (a,b,c), shouldn't the hypotheses reference abnormal returns during the period of the tax policy change
For H2, take out obvious abnormal return and just say abnormal return
For H3, replace correlation with effect
