Table 9 presents the descriptive statistics of the 32 TDRs and the 32 control samples.
Panel A reports the descriptive statistics one year before the TDRs were listed on the TWSE (year -1). In Panel A, the mean assets, sales, and net incomes are approximately
$133.71 billion, $155.13 billion and $12.22 billion for the TDRs, and $212.56 billion,
$206.05 billion and $7.21 billion for the control sample, respectively. The medians of assets, sales, and net incomes are approximately $56.33 billion, $47.88 billion and $3.98 billion for the TDRs, and $63.00 billion, $64.53 billion and $3.6 billion for the control firms, respectively. The mean and median differences between the TDRs and control firms in year -1 are both insignificant, which means that our matching procedure effectively controls for firm size and profitability.
Panel B shows the descriptive statistics in the year that TDRs listed on the TWSE (year 0). In Panel B, we again compare our matching variables between the TDRs and control firms. The results indicate that both the mean and median differences between TDRs and control firms remain insignificant in year 0, which further increases our confidence in the matching procedure. In Panel B, we also summarize the statistics of seven stock performance measures, which are the standard deviation of stock returns, Sharpe ratio, beta, information ratio, Jensen’s alpha, Treynor ratio and abnormal returns.
The mean and median standard deviation of stock returns are 0.187 and 0.157 for the TDRs and 0.120 and 0.116 for their counterparts, respectively. This indicates that TDRs have higher volatility of returns than the control firms, which means that TDRs on average have higher investment risk in the listing year.
The mean and the median Sharpe ratio are -0.350 and -1.063 for the TDRs, respectively, which indicate that the stock returns of TDRs are lower than the free-risk
40
interest rate. The mean and the median Sharpe ratio are 3.494 and 3.274 for the control samples, respectively, which indicate that the stock returns of control firms on average are higher than the free-risk interest rate. Compared with those of the control samples, the results indicate that TDRs have lower excess returns under the same risk in the listing year.
The mean and median beta are -0.079 and 0.002 for the TDRs, respectively, which indicate that the stock returns of TDRs on average have negative correlation with the market in listing year. Conversely, the mean and the median beta are 0.977 and 1.123 for the control firms, respectively, which indicate that the stock returns of control firms on average have a highly positive correlation with the market conditions in year 0.
Compared with those of the control samples, the results indicate that TDRs have a significantly lower correlation with the market in the listing year.
The mean and the median information ratio are -0.520 and -0.429 for the TDRs and -0.102 and 0.537 for their counterparts, respectively, and thus the stock performances of both TDRs and control samples are lower than the industrial average. The control samples have a higher information ratio than TDRs in year 0, although the difference is insignificant.
The mean and the median Jensen’s alpha are 0.020 and -0.012 for the TDRs and are 0.020 and 0.025 for their counterparts, respectively. This indicates that both the TDRs and control firms have similar excess returns in year 0.
The mean and the median Treynor ratio are 0.195 and -0.006 for the TDRs and -0.179 and 0.005 for their counterparts, respectively. However, the positive mean value of 0.195 for the TDRs may not mean that these have a higher risk premium than the control firms in year 0, as this result may be driven by the negative excess returns adjusted by the
41
risk-free rate and negative market beta of TDRs in year 0.
The mean and the median values of abnormal returns are -0.128 and -0.202 for the TDRs and 0.353 and 0.188 for their counterparts, respectively. The TDRs have on average negative abnormal returns in the listing year. In contrast, the control firms have positive abnormal returns in the TDRs’ listing year. Compared with those of the control samples, the results indicate that TDRs have significantly lower abnormal returns in the listing year (-0.481; t=-3.83).
42
TABLE 9 Descriptive Statistics
Panel A: Comparison of TDRs and Control Firms in the Year -1
Means Medians
Variable TDRs Control
Firms
Panel B: Comparison of TDRs and Control Firms in the Event Year (Year 0)
Means Medians
Variable TDRs Control
Firms
*, **, *** Significant at 10%, 5% and 1% levels, respectively, using a two-tailed test.
t-statistic/ z-statistic is for pooled difference of means/Wilcoxon sign-rank test.
Table 9 compares the 32 TDRs with their corresponding 32 control firms. The control firms are matched to ones in the same industry, belonging to the same assets quintile in year -1, and with the closest net income in year -1. Assets are total assets as of fiscal-year end (TEJ data item # 0010), NI are net incomes as of fiscal-year end (TEJ data item # 3950), Sales is net annual sales (TEJ data item # 3100), SD is the standard deviation of stock returns, SR is Sharpe ratio, Beta is market risk (beta), IR is information ratio, Jensen is Jensen's alpha, TR is Treynor ratio, and AR is abnormal returns.
43
Table 10 shows the three-year stock performance of the seven indices for the TDRs after the TDR was listed on the TWSE, as well as those for the matched Taiwanese listed companies. The seven indices include the standard deviation of stock returns, Sharpe ratio, beta, information ratio, Jensen’s alpha, Treynor ratio, and abnormal returns. Among the 32 TDRs, seven were listed in 2009, 11 in 2010, and nine in 2011, and thus the sample of TDRs with a listing period longer than three years is relatively small. We provide details of the TDRs’ three-year stock performance after listing in Table 10, and provide those for the 10-year stock performance after listing in Table E-1 in Appendix E.
The mean standard deviation of stock returns for the TDRs and the control firms are both significantly different from zero through year 0 to year 9, ranging from 0.100 to 0.187 for the TDRs, while from 0.103 to 0.169 for the control firms, with the fluctuations over this period being greater for the former. Comparing their differences by year, we find that TDRs have a significantly higher standard deviation of stock returns in three years after being listed on the TWSE.
Most of the market betas for TDRs are positive throughout the sample years, except for year 0 (-0.079). This negative beta is mainly because almost half the TDR firms are listed after July, and hence do not have sufficient data for this year. The positive beta means that the performance of the TDRs has a positive correlation with that of the market.
For the control sample, the mean values of beta are all positive and remain stable between 0.977 and 1.869, which means that their performance has a positive correlation with the market. The differences between TDRs and control samples are insignificant, except for year 0, which indicates that the changes in the market have a similar influence on both of them.
The Sharpe ratio of the TDRs is negative in most of the years, which means that the stock returns of TDRs on average are lower than the risk-free interest rate. In contrast,
44
most of the Sharpe ratios for the control sample are positive through year 0 to year 9. The TDRs have significantly lower excess returns in comparison to the risk-free interest rate than control samples in year 0 (3.844), year 1 (-1.951), and year 2 (-3.259), respectively.
Most of the information ratios are negative for TDRs, except for year 5 (2.116), which means that on average the TDRs have lower stock returns than the industrial average. In year 1 and year 2, the TDRs have significantly low information ratios (-2.583 for year 1; -2.337 for year 2). The same phenomenon is seen for the control firms, with most of the information ratios being negative, except for years 2, 4 and 6. Compared with their counterparts, the information ratios of the TDRs are significantly lower in year 1 (-1.093) and year 2 (-3.296).
The Jensen’s alphas for TDRs are volatile through year 0 to year 9, ranging from -0.013 to 0.041. The Jensen’s alphas for the control sample remain roughly stable, from -0.008 to 0.030, through year 0 to year 9. The differences in Jensen’s alphas between the TDRs and control sample are statistically insignificant, except for year 1 (0.016) and year 2 (-0.021), which means that the TDRs do not have superior excess stock returns after the listing year.
The Treynor ratios for TDRs are positive in year 0 (0.195) and year 1 (0.061), but then become negative after this. The positive mean value 0.195 of the TDRs does not necessairly mean that that they have a better risk premium than the control firms in year 0, as this may driven by the negative excess returns adjusted by the risk-free rate and negative market beta of TDRs in year 0. The differences between the TDRs and control firms do not show any obvious trend. The difference in Treynor ratios between TDRs and the control sample is - 0.049 in year 2, which suggests that TDRs’ risk premium become lower in the years following this.
45
Most of the abnormal returns for TDRs are negative, with significantly negative abnormal returns in year 0 (-0.127) and year 1 (-0.134). By contrast, on average the abnormal returns for the control sample are positive over the sample period. The differences in abnormal returns between the two samples indicate that TDRs have significantly lower abnormal returns than the control sample in year 0 (-0.480), year 1 (-0.173), and year 2 (-0.274).
We summarize the above results, as follows. In the three years following listing (year 0, 1, 2) on the TWSE, TDRs have a higher risk (standard deviation), lower systematic risk (beta), lower risk-free-rate-adjusted excess returns (Sharpe ratio), lower industry-adjusted excess returns (information ratio), lower excess returns (Jensen’s alpha), lower risk premium in the third year following the listing on the TWSE (Treynor ratio), and lower abnormal returns (CAPM), compared to the control firms. However, over the longer term (year 3 to year 9) the TDRs do not outperform the control firms on the seven indices.
46
TABLE 10
The Indices of Stock Performance for TDRs and Control Samples Following the Listing Years
Indices / Year 0 1 2
*, **, *** Significant at 10%, 5%, and 1% levels, respectively, using a two-tailed test.
Table 10 shows the 3-year performance of the seven indices for the TDRs following the year in which the TDR was listed on the TWSE, as well as those for the matched Taiwanese listed companies. The control firms are matched to ones in the same industry, belonging to the same assets quintile in year -1, and with the closest net income in year -1. SD is standard deviation of stock returns, SR is Sharpe ratio, Beta is market risk (beta), IR is information ratio, Jensen is Jensen’s alpha, TR is Treynor ratio, AR is abnormal returns, and n is the number of test firms.
47
Table 11 reports the three-year stock performance of the seven indices for the full sample following the year in which the TDRs were listed on the TWSE. Among the 36 TDRs, 10 TDRs were listed in 2009, 11 in 2010, and 10 in 2011; as a result, the sample of TDRs with a listed period longer than three years is small. For the full sample, we provide details of the three-year stock performance after listing in Table 11, and 10-year stock performance after listing in Table E-2 of Appendix E.
The standard deviations of stock returns for TDRs and Taiwanese listed companies are both significantly different from zero throughout the sample period. This ranged from 0.100 to 0.212 for TDRs while from 0.130 to 0.195 for Taiwanese listed companies, with a greater fluctuation for the former. Comparing their difference by years, the TDRs have significantly lower standard deviations of stock returns in year 1 (-0.033) and year 2 (-0.027).
Most of the market betas for TDRs are positive throughout the sample years, except for year 0 (-0.148). The negative beta is mainly because almost half the TDR firms were listed after July, and hence do not have sufficient data for this year. In years 1 to 9, the betas for TDRs ranged from 1.054 to 1.811, with the positive beta indicating a positive correlation with the market. For Taiwanese listed companies, the average betas remain roughly stable from 0.717 to 1.171, indicating a positive correlation with the market. The differences between TDRs and Taiwan listed companies are all insignificant, and thus changes in the market have similar influences on both of them.
Most of the Sharpe ratios are negative for TDRs. In particular, TDRs have significantly negative Sharpe ratios in years 1 (-1.593) and 2 (-1.478), which means that on average the stock returns of TDRs are lower than the risk-free interest rate. In contrast, most of the Sharpe ratios for Taiwanese listed companies are positive through years 0 to 9.
Comparing the differences in the Sharpe ratios, the TDRs have lower excess returns in
48
comparison to the risk-free interest rate than Taiwanese listed companies over the sample period, with the differences being significantly negative in years 1 (-1.522) and 2 (-1.873).
Most of the information ratios are negative for the TDRs, except for year 5 (2.115), which means that on average TDRs have lower stock returns than the industrial average.
In years 1, 2 and 3, the TDRs have significantly lower information ratios (-2.507 for year 1; -2.331 for year 2; -2.849 for year 3). The same phenomenon is found for the Taiwanese listed companies, as most of the information ratios are negative, except for year 0 (0.111).
Compared with their counterparts, the information ratios of TDRs are significantly lower than those of the Taiwanese listed companies in years 1 (-1.459), 2 (-1.35), 3 (-2.133), and 4 (-3.412).
The Jensen’s alpha for TDRs is positive in year 0 (0.021). However, it becomes negative in years 1 to 4, ranging from -0.020 to -0.007. This means that TDRs have positive excess returns in the listing year, with the excess returns becoming negative thereafter. The highest Jensen’s alpha for the Taiwanese listed companies appears in year 0 (0.031), and then remains roughly stable from 0.002 to 0.011 through years 1 to 9. The difference in Jensen’s alpha between TDRs and Taiwanese listed companies shows that on average the former have lower but statistically insignificant excess returns than the latter.
The highest mean value of the Treynor ratio is in year 0, at 0.180 for the TDRs.
However, this does not necessaily mean that TDRs have a better risk premium than the market in year 0, as this result may driven by the negative excess returns adjusted by risk-free rate and nagetive market beta of TDRs in year 0. The Treynor ratios then become negative after year 1. The Treynor ratios for the Taiwanese listed companies are very volatile, ranging from -5.553 to 0.145, fluctuating much than is seen for the TDRs
49
throughout the sample period. The differences in Treynor ratios between TDRs and Taiwan listed companies do not show any obvious trend, as these are positive in year 0 (0.388) and year 1 (0.074), but becomes negative in year 2 (-0.123), which suggests that TDRs’ risk premium become worse in the years following year 2.
Most of the abnormal returns for TDRs are negative. Especially, TDRs have significantly negative abnormal returns in year 0 (-0.173), year 1 (-0.114), and year 3 (-0.441). The abnormal returns for Taiwan listed companies are all positive and significant. The difference in the abnormal returns shows that TDRs have significantly lower abnormal returns than Taiwan listed companies, particularly in year 0 (-0.303).
We summarize the above results. In the three years following the listing on TWSE, TDRs have lower risks (standard deviation), don’t have higher systematic risk (beta), have lower risk-free-rate-adjusted excess returns (Sharpe ratio), have lower industry-adjusted excess returns (Information ratio), have lower excess returns (Jensen’s alpha), have lower risk premium in the third year following the listing on TWSE (Treynor ratio), have lower abnormal returns (CAPM), compared to Taiwan listed companies. In the longer period (year 3 to year 9), TDRs do not outperform Taiwan listed companies on the seven indices.
Most of the seven indicators in Table 11 exhibit similar results with those in Table 10 except for the standard deviation of stock returns. Combined with the results in Table 10 and Table 11, the inconsistent findings of the standard deviation indicate that the standard deviation of TDRs is higher than the market, but lower than the matched firms which in the same industry, the same assets quintile, and with similar net incomes.
50
TABLE 11
The Indices of Stock Performance for Full Samples Following the Listing Years
Indices / Year 0 1 2
*, **, *** Significant at 10%, 5%, and 1% levels, respectively, using a two-tailed test.
Table 11 displays 3-year performance of the seven indices for full samples of TDRs and Taiwan listed companies, following the year which the TDR was listed on TWSE. SD is standard deviation of stock returns, SR is Sharpe ratio, Beta is market risk (beta), IR is information ratio, Jensen is Jensen’s alpha, TR is Treynor ratio, AR is abnormal returns. n is the number of test firms.
51
Table 12 displays mean values of ROA, ROE, and discretionary accruals for the TDRs and matched samples in the years around TDRs’ listing on the TWSE. The former two indices measure operating performance, and the latter indicator measures the level of earnings management. We focus on the period from year -2 to year 2 (year 0 is the TDRs’
listing year). In addition, we shed light on two years before (year -1 and year 0) and after (year 1 and year 2) the listing.
The mean values of ROA for the TDRs in the year -2, -1, 0, 1, and 2 are 13.9%, 10.7%, 8.8%, 6.5%, and -0.4%. The TDRs’ ROAs decline year by year. The difference in ROA between the pre- and the post-listing periods for the TDRs is -6.1% (t = -3.33) and is statistically significant, which indicates that TDRs exhibit lower level of efficiency at using their assets to generate earnings following the listing. Relatively, the ROAs for the control firms maintain stably (6.8% in year -2; 7.8% in year -1; 6.8% in year 0; 6.1% in year 1; 6.5% in year 2). The high level of TDRs’ ROAs compared to the control firms in the pre-listing period might be driven by managers’ strong incentives to manage earnings for secondary listing. The difference of ROA between the TDRs and the control sample is significant in the year -2 (7.1%), year -1 (2.9%) and year 2 (-1.8%).The results also show that the TDRs have higher ROAs than the control firms prior to the listing (2.5%; t = 2.36). However, the ROA of TDRs decrease after the listing (-2.7%; t = -1.44).
The mean values of ROE for the TDRs in the year -2, -1, 0, 1, and 2 are 21.8%, 18.3%, 14.7%, 11.7%, and 1.4%, respectively. The TDRs’ ROEs reduce year by year. The TDRs’ ROE substantially drops by 9.1% (t = -3.01) in the post-listing period, which indicates that TDRs exhibit lower level of efficiency at using stockholders’ fund to generate earnings following the listing. By contrast, the ROEs for the control firms maintain stably throughout the years (11.3% in year -2; 15.6% in year -1; 12.0% in year 0;
10.1% in year 1; 11.1% in year 2). The high level of TDRs’ ROEs compared to the
52
control firms in the pre-listing period might be driven by managers’ strong incentives to manage earnings for secondary listing. The difference of ROE between the TDRs and the control sample is significant in the year -2 (10.5%).The results also show that the TDRs have higher ROEs than the control firms prior to the listing (2.7%; t = 1.23). Nevertheless, the ROE of TDRs turns to slightly lower than the control firms after the listing (-3.1%; t
= -1.01).
The discretionary accruals for the TDRs in the year -2, -1, 0, 1, and 2 are 0.129, 0.257, -0.058, -0.347, and 0.075, respectively. The difference in discretionary accruals between the pre- and the post-listing period for the TDRs is statistically insignificant, which suggests that there is no supportive evidence on the attempts of TDRs’ managers to manage earnings with discretionary accruals around the listing. The discretionary accruals for the control firms are 0.247 in year -2, 0.108 in year -1, -0.105 in year 0, 0.305 in year 1, and -0.211 in year 2, respectively. There is also no significant difference
The discretionary accruals for the TDRs in the year -2, -1, 0, 1, and 2 are 0.129, 0.257, -0.058, -0.347, and 0.075, respectively. The difference in discretionary accruals between the pre- and the post-listing period for the TDRs is statistically insignificant, which suggests that there is no supportive evidence on the attempts of TDRs’ managers to manage earnings with discretionary accruals around the listing. The discretionary accruals for the control firms are 0.247 in year -2, 0.108 in year -1, -0.105 in year 0, 0.305 in year 1, and -0.211 in year 2, respectively. There is also no significant difference