Review of Trading on Canadian Equity Marketplaces of Inter-listed Securities Subject to the US Short Sale Circuit Breaker

Transcription

1 Review of Trading on Canadian Equity Marketplaces of Inter-listed Securities Subject to the US Short Sale Circuit Breaker Trading Review and Analysis Investment Industry Regulatory Organization of Canada, Suite 900, 145 King Street West, Toronto, Ontario. M5H 1J8

2 Review of Trading on Canadian Equity Marketplaces of Inter-listed Securities Subject to the US Short Sale Circuit Breaker Study Period February 28, 2011 to April 29, 2011 Further Information Questions or comments regarding this study should be directed to: Victoria Pinnington Vice President, Trading Review & Analysis Investment Industry Regulatory Organization of Canada, Suite King Street West Toronto, Ontario. M5H 3T9 Telephone: Fax: Kevin McCoy, Director, Analytics Investment Industry Regulatory Organization of Canada, Suite 2000, 121 King Street West, Toronto, Ontario. M5H 3T9 Telephone: Fax: i

3 STUDY PERIOD FEBRUARY 28, 2011 TO APRIL 29, 2011 TABLE OF CONTENTS Contents I. Executive Summary... I-1 II. III. IV. Introduction... II-3 Methodology... III-4 Analysis of Short SCB Events... IV-6 1. Overview of Short SCB Events... IV-6 2. Volatility and Short SCB Triggers... IV-7 3. Overall Impact of Short SCB Events on Short Selling... IV-8 4. Short SCB Triggers Analysis of News... IV Selection of Individual Short SCB Events for Further Review... IV Trade Analysis of Individual Short SCB Events... IV-15 V. Conclusion... V-20 ii

4 REVIEW OF TRADING ON CANADIAN EQUITY MARKETPLACES OF INTER-LISTED SECURITIES SUBJECT TO THE US SHORT SALE CIRCUIT BREAKER STUDY PERIOD FEBRUARY 28, 2011 TO APRIL 29, 2011 I. Executive Summary In March 2011, IIROC Market Regulation staff initiated a project to monitor and review the short selling activity on Canadian marketplaces in inter-listed securities for which a Short Sale Circuit Breaker ( Short SCB ) had been triggered in the US for the period February 28, 2011 to April 29, 2011 ( the review period ). The intention of the project was to determine if there was any systemic redirection of short selling activity to Canadian marketplaces, which would normally have taken place exclusively on a US marketplace. The project compared patterns of short selling activity immediately before, during, and after the triggering of the Short SCB for individual securities. During the review period a total of 112 Short SCBs involving a Canadian inter-listed security were triggered. When the data for the 112 Short SCB events was consolidated, we found that with respect to the securities subject to the Short SCB events, the percentage of short selling relative to total volume of trading during the circuit breakers was within the range observed both before and after the period when the circuit breakers were in effect. I-1

5 When each of the 112 Short SCB events was reviewed individually: 50 events (44.6%) had increased levels of short selling during the two-day circuit breaker period when compared to the prior five days, and 62 events (55.4%) had lower levels of short selling, and 52 events (46.4%) had increased levels of short selling during the two-day circuit breaker period when compared to the following five days, and 60 events (53.6%) had lower levels of short selling. A total of 22 Short SCB events were identified in which short selling increased by more than 10% as a percentage of total traded when compared to the pre-scb period. Of these, 13 Short SCB events which exhibited the most significant increase in short selling were reviewed in detail. It was found that in the majority of the events the increase in short selling as an overall un-weighted percentage was attributed to the trading activities of a few trading entities. Although some of the trading entities showed elevated levels of short selling during the circuit breaker periods, the trading activity was representative of their normal trading patterns when the Short SCB restrictions were not in effect and did not appear to be a deliberate redirection of trading that would normally have taken place exclusively on a US marketplace. Based on the analysis of overall levels of short selling, and the in-depth analysis of 13 Short SCB events which exhibited the most significant increase in short selling, IIROC does not believe there is evidence of a systemic migration of short selling activity of inter listed securities to Canadian markets when short sale circuit breakers were in effect in the US. I-2

6 II. Introduction The Short SCB was part of a number of amendments to Regulation SHO approved by the SEC and came into effect on February 28, 2011 as Rule 201. Under the rule, should the price of a security decline at least 10% from the closing price on the primary listing market on the previous trading day, a circuit breaker would be triggered and any short sale executed during the balance of that trading day and the next trading day would have to be entered at a price which was at least one trading increment above the current national best bid. Concerns were raised that the triggering of a Short SCB on an inter-listed security could result in a migration of short selling activity to Canadian markets, which do not have any short sale price restrictions on inter-listed securities. This review was conducted to determine whether any significant increases in short sale activity in Canada while a Short SCB was invoked in the US could be identified and whether any increase appeared to be caused by a migration of short sale order flow from the US markets. This project reviewed the short selling activity in inter-listed securities on Canadian marketplaces for which a Short SCB was invoked during the period of February 28, 2011 to April 29, 2011 (the review period ). The review compared patterns of short selling activity immediately before, during and after the triggering of the Short SCB. II-3

7 III. Methodology For the purposes of the study, inter-listed is defined as any security which is listed on both a Canadian equity marketplace and a recognized US equity marketplace 1. The Short SCB alerts originating from the US were monitored and alerts involving interlisted securities were identified. For each alert involving an inter-listed security the short sale volume as a percentage of total volume was reviewed over a period of twelve business days, broken into three sub-periods, including; the five business days prior to the triggering of the Short SCB ( Pre-CB Period ), the two-day restricted period of the Short SCB ( Circuit Breaker Period ), and the five business days following the triggering of the Short SCB ( Post-CB Period ). Over the review period, a total of 112 Short SCBs were identified where the relevant security was inter-listed on a Canadian exchange. Securities which showed an increase in short selling activity as a percentage of total trading of 10% or more from the Pre-CB Period to the Circuit Breaker Period were selected 2 for further and more detailed review. A total of 22 of the 112 met the criteria for further review. The more detailed review included analysis of the trading during the three sub-periods, an analysis of the trading by trading entities who were short selling during and around the Circuit Breaker Period, and any relevant events that may have affected the price of 1 The definition of recognized US equity marketplace applied here is that which aligns with the definition of foreign organized regulated market in UMIR % is not statistically significant and was used only as a method of selecting a relevant sample for further review. The percentage change was security specific relative to day over day activity and does not consider general market movements, volatility or any other market event. III-4

8 the security such as company news, world events, or overall market trends. Where warranted, the trading activity that occurred on the triggering day was separated into pre-trigger and post-trigger groups to differentiate between trading that occurred while the restriction was in effect and when it was not. 3 This facilitated a more detailed analysis of the actual effect of the Short SCB on short selling activity. Where necessary the identity of the entity engaged in short selling activity was further clarified to ascertain if it was a US-based client. 3 The provisions of Rule 201 provide that the triggering of a SCB event may occur any time during the trading day. Accordingly, the restrictions may only be applicable to the latter part of the trading day in which the event is triggered. III-5

9 IV. Analysis of Short SCB Events 1. Overview of Short SCB Events During the Review Period, a total of 112 Short SCB events that involved a Canadian inter-listed security were identified. Below is a summary of the listing market, frequency and timing: TSX TSXV Total Short SCB Alerts # of Securities Securities with One Short SCB Two Short SCBs Three Short SCBs Four Short SCBs Five Short SCBs Time of Short SCB: 9:30-10: :01-12: :01-17: IV-6

10 2. Volatility and Short SCB Triggers Figure 1 The chart in Figure 1 plots the relationship between the frequency of the U.S. circuit breaker triggers and the volatility of the overall market and illustrates that for the most part, the frequency of the triggers parallels the volatility of the market 4. 4 Market Volatility was measured using the daily high and low of the S&P TSX composite Index IV-7

11 3. Overall Impact of Short SCB Events on Short Selling Figure 2 Figures 2 and 3 illustrate the overall market impact of the Short SCB events on short selling activity on Canadian markets. In Figure 2, the percentage of short sales relative to the total volume of trading is shown for each day in the three time periods. All 112 Short SCB events are included, and when events overlapped in the same security, each event was accounted for separately. From this chart we can see that overall, the level of short selling in the Circuit Breaker Period (indicated by a slightly darker colour) was consistent with the level of short selling in the Pre- and Post-CB Periods. IV-8

12 Figure 3 In Figure 3, the volume of short sales relative to the total volume of trading is shown for each day. From this chart we can see a large spike in the total trading volume on day 1 of the Circuit Breaker Period. It should be noted that day 1 of the Circuit Breaker Period includes intraday trading before and after the Short SCB was active. The volume of trading shows a general decrease in the days following a Circuit Breaker Period. IV-9

13 4. Short SCB Triggers Analysis of News Figure 4 Figure 4 shows that 83% of Short SCBs were triggered when there was news. In 53% of the cases (blue bar), the news concerned the overall market condition or specific sector news. In 30% of the cases (dark green bar), the news was directly relevant to the specific security. The company specific news has been broken out by type of news (light green bars). IV-10

14 5. Selection of Individual Short SCB Events for Further Review Of the 112 Short SCB Triggers, 21 (or 18.75%) TSX-listed securities and 1 (or.89%) TSX V-listed security experienced an increase in short volume percentage greater than 10% during the Circuit Breaker Period when compared to the Pre-CB Period. Figure 5 Figure 5 illustrates the selection criteria. The red dots represent the average short selling volume as a percent of the total traded volume in the five-day Pre-CB Period for each Short SCB event. A corresponding blue dot either directly above or directly below the red dot represents the average short selling volume as a percent of the total traded volume in the two-day Circuit Breaker Period. The dark blue line represents the 10% threshold line above the Pre-CB Period which was used to determine which securities IV-11

15 would be reviewed in further depth. From this chart it can be seen that in approximately half of the Short SCB events, the percent of short selling was lower during the Circuit Breaker Period compared to the Pre-CB Period. The initial analysis of the individual 22 Short SCB events which demonstrated an increase in short selling of greater than 10% in the Circuit Breaker period revealed that several of the events could be excluded from further review based on the following factors: Recalculation When the short selling activity which occurred on the day of the Short SCB trigger was divided into pre-trigger and post-trigger volumes, the post-trigger short selling volume fell well below the 10% increase threshold for review. In these cases, the increased amount of short selling took place prior to the Short SCB trigger when no restrictions were in effect. As a result of this recalculation, no further review was conducted for five of the Short SCB events. Illiquidity When the short selling activity during the Circuit Breaker Period was reviewed it was noted that for one event the increased percentage of short selling was attributable to only one trade. As the total trading for that day was extremely (and uncharacteristically) low, the percentage of short selling during the Short SCB was deceptively high. As a result of this observation, no further review was conducted for one of the Short SCB events. Trend of Short Selling / News Triggers Fall in Price In several of the selected Short SCB events, it was noted that the percentage of short selling remained as high or higher in the Post-CB Period than in the Circuit Breaker Period. In these cases, the short selling during the Circuit Breaker Period would appear IV-12

16 to be in line with the Post-CB Period trading. As an example, if the price drop and resulting Short SCB trigger occurred because of a significant material event, such as the Japanese earthquake, the resulting increase in short selling observed in Canadian marketplaces may have been a natural consequence which began after the news and continued well after the end of the Short SCB, and not representative of a migration of short selling to Canadian marketplaces. In order to explain this concept fully, Figure 6 illustrates the four scenarios for all 112 of the Short SCB events, considering the patterns of short selling both prior to and following the Circuit Breaker Period. Figure 6 i. 45 of the Short SCB events showed a pattern in which the percentage of short selling fell during the Circuit Breaker Period and rose in the Post-CB Period, forming IV-13

17 a trough in short selling. These events are in the lower left hand quadrant of Figure 6. ii. iii. iv. 17 of the Short SCB events showed a pattern in which the percentage of short selling fell from the Pre-CB Period to the Circuit Breaker Period to the Post-CB Period. These events are in the upper left hand quadrant of Figure of the Short SCB events showed a pattern in which the percentage of short selling rose from the Pre-CB Period to the Circuit Breaker Period and continued to rise through the Post-CB Period. These events are in the lower right hand quandrant of Figure of the Short SCB events showed a pattern in which the percentage of short selling peaked in the Circuit Breaker Period relative to the Pre- and Post-CB Periods. These events are in the upper right hand quadrant of Figure 6. The events to the right of the red line (in quadrants iii and iv) are the 22 events which were originally selected as the result of a 10% increase in short selling between the Pre- CB Period and the Circuit Breaker Period. The yellow box in the upper right quadrant highlights the events for which the short selling activity as a percentage of total trading in the Circuit Breaker Period was also 10% or more than the Post-CB Period. An analysis was conducted of the news relevant to the securities for the events which fell outside of the yellow box (to the right of the red line). Where there was significant news and a continued upward trend of short selling, the event was excluded from further review. As a result of this analysis, no further review was conducted for three of the Short SCB events. IV-14

18 6. Trade Analysis of Individual Short SCB Events As detailed above, a total of 9 Short SCB events were excluded from further analysis, leaving 13 of the Short SCB events to be reviewed in detail. The entities responsible for more than 10% of the total short sale volume in the Circuit Breaker Period were identified, and their trading activity during the whole review period was charted, both by the absolute buy/sell/short sell volume, and by the percentage buy/sell/short sell volume. The trading activity of each entity was reviewed to determine if it was in context with the market (showing high trading volumes on the days of overall high trading volumes) and if the activity during the Circuit Breaker Period days was consistent with the pattern of trading in the Pre- and Post-CB Periods. For each event reviewed, it was found that two or three entities accounted for 10% or more of the short selling during the Circuit Breaker Period. In total, 8 clients accessing the markets through direct market access ( DMA clients ), 3 Trader IDs trading for inventory and two Trader IDs trading as agent for clients were responsible for short selling more than 10% in the 13 Short SCB events. The 8 DMA clients have been divided into three categories, based on their nature of trading, as follows: Five out of eight of the DMA clients showed a pattern of buying and selling a security in unequal amounts on a daily basis, thus appearing directional on a daily basis. We have identified these clients as directional DMA clients ( DDMA clients ). Three out of eight of the DMA clients showed a pattern of buying and selling a security in equal amounts on a daily basis, thus appearing non-directional. We have called these clients non-directional DMA clients ( NDDMA clients ). IV-15

19 Figure 7 Unweighted Percentage of Short Selling Nature of Trading 57% Directional DMA Clients 25% Not Reviewed - Less than 10% of Short Selling in Event 7% Client Trading 7% Non-Directional DMA Clients 4% Inventory Trading Figure 7 shows the unweighted percentage of short selling in the thirteen events categorized by the nature of trading. To create this table, the percentages of short sale volumes for each Short SCB event for each category were added together and divided by 13. As such, the results do not take into account the differences in trading volumes or values. In layman s terms, we are comparing apples to oranges. Nevertheless, this table is presented in order to provide an overall summary of the short sale trading activity during the Circuit Breaker Period in the reviewed securities in order to assess the extent of potential migration of short selling activity to Canadian markets. The following observations and conclusions were made regarding short selling during the Circuit Breaker Periods: Directional DMA Clients In twelve out of the thirteen events reviewed in detail, accounting for approximately 57% of all the short selling, DDMA clients were responsible for significant percentages of short selling during the Circuit Breaker Periods. After further analysis of each individual event, the following observations were made: IV-16

20 The DDMA clients were buying and selling the securities on almost all of the twelve days of the review periods. The DDMA clients sold short on the Canadian marketplaces in the Pre-CB and Post-CB Periods, some days at percentages matching or exceeding the percentages of short selling on the Circuit Breaker days. In seven of the thirteen Short SCB events, one or more of the DDMA clients sold short more than twice the percentage by volume during the Circuit Breaker Period compared to the Pre-CB Period. In seven of the thirteen Short SCB events, one or more of the DDMA entities sold short more than twice the percentage by volume during the Circuit Breaker Period compared to the Post-CB Period. The following factors should be considered when evaluating the significance of this trading activity: The specific DDMA client trading strategies are not known, but it is known that the DDMA client strategies were executed using algorithms. The short selling seen in the Pre-CB and Post-CB Periods supports the theory that short selling is a fundamental part of the way that these DDMA clients conduct their trading on a daily basis. The mechanism by which the individual DDMA clients modified their trading execution methods to comply with the US short sale restriction is not known, but it is reasonable to assume that they did so. Based on the evidence reviewed and the additional factors explored, it would appear that the trading activity by the DDMA clients during the Circuit Breaker Periods showed a logical and expected increase in short selling on Canadian marketplaces. It would appear that the increase in short selling during the Circuit Breaker period in the reviewed securities by the DDMA clients was representative of their normal course IV-17

21 trading patterns. The trading by the DDMA clients does not appear to represent a deliberate redirection of trading which would normally have taken place exclusively on a US marketplace. Not Reviewed - Less than 10% of Short Selling in Event Approximately 25% of the short selling was not reviewed, as it represented less than 10% of the short sale volume during the individual Short SCB events. It should be noted that the DDMA and NDDMA clients already identified were also present in some Short SCB events, short selling less than 10%. Client Trading In one of the Short SCB events reviewed in detail, accounting for approximately 7% of all the short selling, the high volume of short selling was attributable solely to client trading at two Participating Organizations. The client trading was not part of a larger pattern. It is possible that this trading may have been routed to a Canadian market in preference to a US market. However, as this isolated client activity was only attributed to one occurrence, it is not indicative of a potential systemic issue. Non-Directional High Frequency Trading In four out of the thirteen Short SCB events reviewed in detail, accounting for approximately 6% of all the short selling, NDDMA clients were responsible for significant percentages of short selling during the Circuit Breaker Periods. After further analysis of each individual event, the following observations were made: The NDDMA clients were buying and selling the securities on all twelve days of the review periods, and their trading showed equal amounts of buying and selling each day. IV-18

22 The NDDMA clients short sold on the Canadian marketplaces in the Pre-CB and Post-CB Periods, on most days at percentages matching the percentages of short selling on the Circuit Breaker days. In one of the Short SCB events, one NDDMA client ( ND Exception ) sold short more than twice the percentage by volume during the Circuit Breaker Period compared to the Post-CB Period In none of the other Short SCB events did the NDDMA clients sell short significantly more during the Circuit Breaker Period when compared to the Pre- CB or Post-CB Periods. The following factors should be considered when evaluating the significance of the trading activity by the ND Exception: The short selling seen in the Pre-CB and Post-CB Periods supports the theory that short selling is a fundamental part of the way that the ND Exception client conducts its trading on a daily basis. The mechanism by which the ND Exception client modified its trading execution methods to comply with the US short sale restriction is not known, but it is reasonable to assume that it did so. Based on the evidence reviewed and the additional factors explored, it would appear that the trading activity by the ND Exception client during the Circuit Breaker Period showed a logical and not-unexpected increase in short selling on Canadian marketplaces. It would appear that the increase in short selling during the Circuit Breaker period in the reviewed securities by the ND Exception client was representative of their normal course trading pattern. The trading by the ND Exception client does not appear to represent a deliberate redirection of trading which would normally have taken place exclusively on a US marketplace. IV-19

23 The trading activity by the remainder of the NDDMA clients during the Circuit Breaker Periods was representative of their normal course trading patterns, and did not show any evidence of a migration of short selling activity to Canadian markets. Inventory Trading In three out of the thirteen Short SCB events reviewed in detail, accounting for approximately 4% of the short selling, inventory trading by Canadian firms was responsible for significant percentages of short selling in the Circuit Breaker Periods. The inventory accounts are Canadian, and it is unlikely that the trading activity represents migration of short selling activity to Canadian markets. V. Conclusion When the 112 Short SCB events are reviewed as a group (see Figure 2), we find that the percentage of short selling relative to total volume of trading ranges from 28% to 34% over the twelve-day period from Pre- to Post-CB Period. Day One of the Circuit Breaker period exhibited an average rate of short selling of 33%, and Day Two exhibited an average rate of short selling of 31%, both of which are within the average range for the twelve-day period. This shows that on average, there was no increase in the rate of short selling on Canadian marketplaces during the Short SCB events. Figure 5 supports this finding, showing that approximately half of the 112 Short SCB events had increased short selling during the Circuit Breaker Period, and half had decreased short selling. A group of 22 Short SCB events that experienced the largest increases in short selling was reviewed. An initial review resulted in excluding nine of the events from further review. The remaining 13 Short SCB events were subjected to an in-depth review of trading to identify the entities responsible for the short selling, and to determine if this short selling was in character with the entity s typical and expected trading behaviour V-20

24 or if it appeared to be a deliberate redirection of trading which would normally have taken place exclusively on a US marketplace. The increase in short selling in the majority of the events and as an overall un-weighted percentage (Figure 7) was attributed to the trading activities of a few DDMA clients. Although the DDMA clients showed elevated levels of short selling during the Circuit Breaker Periods, the trading activity was representative of their normal course trading patterns when Short SCB restrictions were not in effect and did not appear to be a deliberate redirection of trading which would normally have taken place exclusively on a US marketplace. In conclusion, IIROC does not believe there is evidence of a systemic migration of short selling activity of inter-listed securities to Canadian markets when Short SCBs were in effect in the US. V-21