We first report on the measure of conflict of interest that occurs in investment banks and then report on the measure of conflict of interest that occurs in stocks.
Tables 3 to 8 present the basic statistics, including the mean, minimum and
maximum of the three measures of conflict of interest: the conflict of interest byweeks (CI_W), the conflict of interest by amounts (CI_A), and the conflict of interest by shares (CI_S
N) across all recommendations issued of each of the security houses for each year, and recall that the mean values of these measures are also used to measure the conflict of interest index for investment banks: CI_IB_W, CI_IB_A, and CI_IB_S, respectively. A positive number is indicative of a positive
net sell, and therefore, the presence of conflicts of interest is implied. The higher the positive number is, the more pervasive are the conflicts of interest. Banks arelisted in the tables in descending order on the basis of the mean values from positive to zero and to negative. The term na means that that security house does
not make any buy recommendations during that year.In the columns on the far left hand side of Table 3, we take the year 2000 as the example to illustrate the extent to which each security house is involved in conflicts of interest using CI _W1. There are 13 security houses that obtain positive average scores, 6 that obtain zero and 4 that obtain negative average scores. As shown, the top ranking security house N has the highest average scores as it makes
10
68 recommendations, and the maximum and minimum value or measure of conflict of interest is respectively +2 and –2. A maximum of +2 means that in a particular week, the security house sells each buy recommended stock one week both before and one week after it issues that recommendation; in this way, it receives the highest score when it is involved in a conflict of interest. When it is –2, it is in no conflict of interest whatsoever since the security house also buys the buy recommended stock.
In the case of security house N, the average score is 0.294, a sign that the investment bank sells each recommended stock, on average, for 0.294 weeks out of the possible two weeks. The interpretation of this, of course, is that the security house is occasionally, but certainly not always, involved in a conflict of interest.
Several other interesting findings are worth noting for the year 2000. For one, the number of recommendations differs substantially across security houses.
Security house R, for example, makes 87 recommendations, while H, W, M, V, III and A make none. Simply judging from the figures, all in all, a higher number of recommendations tends to be associated with a more positive average measure of conflict of interest. In addition, eleven security houses have a maximum of 2, a strong indication that at no time have they wanted to hold the stocks they have recommended; on the contrary, they have wanted to sell them. What’s more, seven security houses have maximum scores of zero, indicating that they have not taken any further action with a particular stock after having recommended it.
Similar findings, albeit with slight differences, are shown between 2001 to 2003. First, it is noted that the number of security houses with a positive score or average measure of conflict of interest steadily drops from 13 to 11 in 2001, to 7 in 2002 and to 5 in 2003. Thus, the prevalence of conflicts of interest is less and less pronounced each year. One reason for this gradual decrease may be that supervision or governance improved during those years. The decrease could also be attributed to worsening economic conditions throughout those three years, which could have made buy recommendations more honest reflecting that security houses might have
been more forthright when making suggestions, or alternatively, in light of the good economic times, investors might have been more prudent and cautious. Next, the number of firms with the maximum score of +2 is, for all intents and purposes, stable throughout the entire four-year period: 11 in 2000, 9 in 2001, 11 in 2002 and 12 in 2003. Evidently, among those firms, the temptation to send the wrong signal does not dissipate. Third, considerably fewer firms make recommendations in 2003, perhaps largely because an increase in the number of mergers, acquisitions and control may have decreased the number of security houses or because 2003 marked the year with the worst economic conditions within the past three decades in Taiwan.
Table 4 presents the same measures of conflict of interest except that we use
a two-week window. The maximum and minimum value respectively becomes +4 and –4. We take 2000 as our benchmark; overall, among all of the security houses, 10 have positive average scores, 5 have scores of zero, and 9 have negative average scores. The number with a positive score jumps to 12 in 2001 but dips to 9 in 2002 and suddenly plunges to only 6 in 2003. This signifies that the perceived importance of using conflicts of interest as part of a business strategy appears to have subsided in more recent years. Beyond this, the number of security houses with the maximum of +4, though it fluctuates, does ultimately decrease somewhat from 8 to 6 and then suddenly to just 2 before showing a marked increase back up to 5 in the above consecutive years.Tables 5 and 6 are the same as Tables 3 and 4 but instead of being based on CI_W
1 andCI_W
2 to determine the pervasiveness of conflicts of interest, they are based on CI_A1 and CI_A2, respectively. A positive value suggests the presence of a higher degree of conflict of interest, whereas a negative value suggests a lower degree of conflict of interest. The maximum value means that when a particular security house makes a buy recommendation, the largest number of net sells occurs.Take the year 2000, which uses a one-week window as the example (Table 5).
Security house IV has the largest maximum with total net sells of around $590
million
($590,431,000)
new Taiwan dollars, which is quite inconsistent with the total buy recommendations. Bear in mind, however, that the other security houses also have noticeably high maximum positive amounts; thus, their proprietary trading or dealer divisions must also be bringing in large profits on account of their high measure of conflict of interest. Granted that the frequency of issuing a recommendation does typically appear to be positively correlated with the average measure of conflict of interest in Tables 3 and 4, but this certainly does not seem to be the case with some security houses in Tables 5 and 6. For example, the highest number of recommendations is 87 followed by 81, but the average value or measure of conflict of interest for each firm is negative. Similar to the results in the previous tables (Tables 3 and 4) nevertheless, the positive average scores decrease over the years. Compared with 12 security houses, which have a positive average score in 2000, only 6 do in 2003. This again, signifies that the measure of conflict of interest decreases with time.Table 6 is the same as Table 5 except that we use the two-week period for CI_A
2. There are 11, 11, 10 and 5 security houses with a positive average score in each consecutive year during the four-year period, suggesting that conflicts of interest are at play. Contrast this with the respective number of security houses with zero or negative scores: 13, 10, 13 and 12, which undoubtedly indicates the absence of any conflicts of interest. Not unlike what is observed in Table 5, the number of security houses with a positive average score seems to have been on the decrease over the years, implying that fewer and fewer investment banks are making use of conflicts of interest as a way to make gains.The results shown in Tables 7 and 8 correspond to those in Tables 3 to 6, except that we use total shares as a measure of conflict of interest. The results are similar to those in the other four tables, where the number of shares with a positive value is 11, 13, 8 and 6 for the one-week window and 9, 12, 11 and 5 for the two-week window.