Results: The impact of analysts coverage
6.5 Effects of different type of recommendations
6.5.5 Recommendations effects on illiquidity measures
Previously in Section 6.4.5 we have showed that the mean illiquidity measure λa in the participating group is significantly higher than the control group. It means that the participating companies are less liquid than companies in the control group. In the next analysis, we compare the illiquidity measure according to the analysts recommendations between the buy and hold companies across all sub-periods.
The results in Table 6.29 show that there is no significant difference of λa between different types of recommendations for each sub-period. It means that companies receiv-ing favourable recommendations do not show any significant impact on liquidity from companies that received less favourable recommendations. Although the differences of λa is not significant from zero, the illiquidity measure in companies with the buy recommen-dations is lower than the hold recommenrecommen-dations in the sub-period after the event. This is consistent with the earlier result in Section 6.5.4 in which the bid-ask spread in the buy recommendations is significantly lower than the hold recommendations in the sub-period after the event.
Table 6.29: Illiquidity measure, λa - buy vs. hold recommendations Buy Hold Difference t-value p-value (2-tailed) Before event 1.685 2.732 -1.047 -1.078 0.286
Event day 1.567 0.853 0.714 0.779 0.440
After event 1.659 3.010 -1.351 -1.315 0.194
Our analysis is continued with a one-way ANOVA test in which we assume that λa for each sub-period and each type of recommendation is randomly and independently distributed. The ANOVA test gives F -value and p-value of 1.307 and 0.264 respectively, which means that there is no group that has significantly higher λa than the others.
The ANOVA test confirms that there is no statistical significant difference of illiquidity measure between each type of recommendations.
Essentially, there is no clear evidence that the illiquidity measure, λa, manages to show that the analysts favourable recommendations have significant effects in reducing illiquidity on the event day. The results reveal that the analysts recommendations are not so influential and informative that they could affect the illiquidity measure.
We have also shown in Section 6.4.5 that our proposed illiquidity measure λb in the participating group is significantly higher than the control group for the sub-period be-fore the event. Our next investigation is to compare the alternative illiquidity measure λb according to the type of analysts recommendations. Table 6.30 compares the mean illiquidity measure between the buy and hold companies across sub-periods. The results
are consistent in which there is a significant difference of λb between the buy and hold companies at the 10% significant level in the sub-period before the event. It means that before the event companies with the buy recommendations are more liquid than compa-nies with the hold recommendations. However, after the event the significant difference of λb disappears. This implies that the analysts favourable recommendations have no significant impact on liquidity.
The results using λb as shown in Table 6.30 is slightly different from the results using λa as shown in Table 6.29. Nevertheless, the pattern of magnitude of illiquidity mea-sures is similar such that companies with the buy recommendations are more liquid than companies with the hold recommendations.
Table 6.30: Alternative illiquidity measures, λb - buy vs. hold recommendations Buy Hold Difference t-value p-value (2-tailed)
Before event 166 375 -209* -1.851 0.070
Event day 142 111 31 0.308 0.760
After event 181 353 -172 -1.350 0.183
notes: * significant at 10%
Similar with the previous analysis, by assuming that each group is randomly and independently distributed, we then apply the one-way ANOVA test which gives F -value of 1.898 and p-value of 0.097. It means that the alternative illiquidity measure between the groups across recommendations and sub-periods is statistically significant at the 10%
level. The result is not consistent with the earlier one using λa. This means that one of the groups in the buy or hold recommendations has significantly higher or lower λb than the rest of the groups.
The multiple mean comparison procedures using the LSD procedure confirms within the hold companies that the difference of λb between sub-periods is statistically significant at the 5% level. This is because λb for the hold recommendations on the event day is very low. Since the calculation of event day illiquidity measure is biased when the company is not traded, Amihud (2002) suggests the using of more observation points in the calculation of illiquidity measure. Accordingly, the calculation of λa and λb on
the event day is not very useful when the companies are not traded on the event day.
Therefore, the using of single observation to calculate illiquidity measures on the event day does not reveal comparable results.
Nevertheless, we have used Amihud (2002)’s illiquidity measure, λa and our own alter-native illiquidity measure, λb. Both measures of illiquidity using the sub-period 120-day before and 120-day after the event reveal similar results in terms of patterns of magnitude of illiquidity measures. Although the participating group has higher illiquidity measures than the control group, the differences within the group itself are not significant from zero. The difference of illiquidity measures between the sub-period 120-day before the event and after the event within each group is also not significant from zero.
In general, both illiquidity measures show the same pattern of results, and therefore the proposed illiquidity measure λb could still be used. Basically, both illiquidity measures reveal that the participating group is less liquid than the control group even though the difference is not statistically significant from zero in most of the tests. Despite the difference in composition of companies between the participating companies and the control group, the difference between the two groups is still comparable.
The insignificant difference of both illiquidity measures between the sub-period before and after the event is an indication that the analysts’ initial reports are not influential and informative such that the reports do not have significant effects in reducing the risk of trading or increasing the ease of trading. This could be due to the content of the reports which is not highly influential nor important to investors, as such there is no impact on investors’ prior belief and expectation which could reduce the risk. Accordingly, the analysts recommendations do not generate and influence investors to trade more vigorously as there is no need for investors to reshuffle their investment portfolio.
The result implies that the analysts recommendations that are accompanying the ana-lysts reports are also not influential nor informative to investors so there is no significant change between the favourable and less favourable recommendations. As there is no change in investors’ expectation upon the release of the analysts recommendations, even the favourable recommendations are not able to attract investors to buy more shares or increase their trading.