What Happened

NYSE suspended trading in all symbols on the exchange at 11:32 am on July 8, 2015, and reopened at 3:10 pm - a suspension of trading for 3 hours and 38 minutes.[1] The following is an analysis of trading in corporate stocks before, during and after the NYSE trading suspension.[2]

As discussed below, total trading volume in NYSE-listed corporate stocks on July 8 was well within the range of observed trading volume in the first seven months of 2015. While the NYSE’s share of trading volume declined, other market centers - especially other exchanges - experienced unusually large increases in trading volume in NYSE-listed stocks on July 8.

During the NYSE trading suspension, on average, NYSE-listed stocks experienced substantial increases in spreads and substantial decreases in depths. Not all NYSE-listed stocks were equally affected by the NYSE trading suspension. Large cap NYSE-listed stocks experienced the biggest declines in depth, while small cap NYSE-listed stocks experienced the biggest increases in spreads.

Impact on Trading Volumes

Figure 1 shows the evolution of trading volume for NYSE-listed (red line) and Nasdaq-listed (orange line) stocks during January to July, 2015. Dollar volume traded for NYSE-listed stocks during normal market hours on July 8 (green vertical line) was $87.3 billion, well within the range of $65 billion to $126 billion during January to July, 2015.[3] For Nasdaq-listed stocks, $55 billion traded on July 8 and the year’s range was $39 billion to $75 billion. The ratio of NYSE-listed volume to Nasdaq-listed volume (blue line, right scale) on July 8 is also well within the range.

Figure 1: Daily Dollar Trading Volume by Listing Exchange – January to July, 2015

Figure 2 shows the distribution of trading volume from January to July 2015 for each of the three intraday time periods defined by the July 8, 2015 NYSE trading suspension: the “pre-event” window (9:30 am to 11:32 am), the trading suspension “event” window (11:32 am to 3:10 pm), and the resumption of trading or the “post-event” window (3:10 pm to 4:00 pm). There is no conclusive evidence that the distribution of trading volume across the day on July 8 was unusual, as compared to other days from January to July, 2015.

Figure 2: Daily Percentage of NYSE-listed Shares Traded – January to July, 2015

Impact on Market Share

If overall NYSE-listed trading volume on July 8 was similar to trading volume on other days, and trading volume did not shift into the pre- or post- NYSE trading suspension period, then where did it go? Figure 3 shows that NYSE’s share of trading in NYSE-listed stocks during the pre-event window (9:30 am to 11:32 am) was lower on July 8 than on the surrounding days. Figure 4 shows that trading in NYSE-listed stocks (red line) shifted to other exchanges during the trading suspension event window (11:32 am to 3:10 pm) on July 8, with Nasdaq (orange line) attracting a large proportion of the trading in NYSE-listed stocks and many other exchanges, particularly Arca, BATS-Z, and Edge-X, picking up market share as well. The TRF (blue line) share was not much different than other days in 2015. Figure 5 shows that NYSE’s share of trading during the post-event window (3:10 pm to 4:00 pm) continued to suffer after the resumption of trading. In short, it appears that traders migrated to other exchanges when the NYSE temporarily suspended trading on July 8, 2015.

Figure 3: Daily Market Share by Market Center During the Pre-Event Window (9:30 am – 11:32 am) – April to July, 2015

Figure 4: Daily Market Share by Market Center During the Event Window (11:32 am – 3:10 pm) – April to July, 2015

Figure 5: Daily Market Share by Market Center During the Post-Event Window (3:10 pm – 4:00 pm) – April to July, 2015

Impact on Spreads

Figure 6 shows the evolution of equal-weighted mean relative quoted spreads for corporate stocks on July 8, 2015.[4] NYSE-listed stocks (red line – left axis) experienced a noticeable increase in spreads in the ten minutes bracketing the 11:32 am NYSE suspension (green vertical line). From 11:24 am to 11:43 am, the equal-weighted mean relative quoted spread for NYSE-listed stocks increased from 18.3 basis points to 26.9 basis points - a 47% increase. NYSE-listed spreads did not reach their pre-11:32 am levels until after the resumption of trading on NYSE at 3:10 pm. In contrast, Nasdaq-listed stocks (orange line – left axis) experienced a modest increase in spreads. The equal-weighted mean relative quoted spread for Nasdaq-listed stocks during the 11:24 am to 11:43 am time interval increased from 1.17% to 1.19% – a 1.7% increase – and reached new lows by noon. As a result, the ratio of mean spreads (blue line – right axis) shows that NYSE-listed stocks experienced a greater impact than Nasdaq-listed stocks during the event window, and especially during the first 75 minutes after trading suspension.

Figure 6: Equal-weighted Mean Quoted Spreads (%) for All Stocks on July 8, 2015

The mean relative quoted spreads portrayed in Figure 6 are equal-weighted averages across all stocks for each one-minute snapshot, computed separately for each listing venue. It is possible that this weighting choice could obscure some feature of large or actively-traded stocks while accentuating some feature of small or thinly-traded stocks.

Figures 7 and 8 provide mean relative quoted spreads weighted by market capitalization and trading volume weights, respectively.[5] Overall, the patterns are similar to those observed with equal-weighted mean spreads. Quoted spreads for NYSE-listed stocks rose immediately prior to the trade suspension and plateaued for the next 75 minutes. The ratio of NYSE-listed spreads to Nasdaq-listed spreads remained elevated throughout the event window. However, the relative change in spreads (blue line) is not as stark in Figures 7 and 8 as in Figure 6, suggesting that smaller and less actively traded NYSE-listed stock spreads might have been impacted more by the trading suspension.

Figure 7: Market Cap-weighted Mean Quoted Spreads (%) for All Stocks on July 8, 2015

Figure 8: Dollar Volume-weighted Mean Quoted Spreads (%) for All Stocks on July 8, 2015

Figures 9, 10 and 11 show mean equal-weighted relative quoted spreads for stocks broken down by market capitalization. Three size bins are created from NYSE-listed market capitalization percentiles: small cap stocks are stocks with market capitalizations between the minimum and 30th percentile of NYSE market capitalizations; mid cap stocks are stocks with market capitalizations between the 30^th and 70^th percentiles of NYSE market capitalizations; and large cap stocks have market capitalizations greater than the 70^th percentile of NYSE market capitalizations. Nasdaq-listed stocks are assigned to bins based on the thresholds established by NYSE market capitalizations.[6]

Figure 9 shows that the mean relative quoted spread for small cap NYSE-listed stocks (the smallest 30 percent of NYSE-listed stocks) rose from 41.9 bp to 63.2 bp (a 51% increase) during the 11:24 am to 11:43 am interval surrounding the NYSE trading suspension and remained elevated for most of the remainder of the trading suspension. For small cap Nasdaq-listed stocks, mean relative quoted spread rose from 1.35% to 1.38% (a 2% increase) over the same interval and tended to decline throughout the remainder of the day.

Figure 9: Equal-weighted Mean Quoted Spread (%) for Small Cap Stocks on July 8, 2015

Figure 10 shows that the mean relative quoted spread for mid cap NYSE-listed stocks (the middle 40 percent of NYSE-listed stocks) rose from 11.4 bp to 15.6 bp (a 37% increase) during the 11:24 am to 11:43 am interval surrounding the NYSE trading suspension, but tended to decline throughout the remainder of the day. Over the same interval, mean relative quoted spread for mid cap Nasdaq-listed stocks rose from 15.7 bp to 17.1 bp (a 9% increase) and also declined throughout the day.

Figure 10: Equal-weighted Mean Quoted Spread (%) for Mid Cap Stocks on July 8, 2015

Figure 11 shows that the mean relative quoted spread for large cap NYSE-listed stocks (the largest 30 percent of NYSE-listed stocks) rose from 4.50 bp to 5.97 bp (a 33% increase) during the 11:24 am to 11:43 am interval surrounding the NYSE trading suspension. Though the time series is noisy throughout the trading suspension, quoted spreads for NYSE-listed stocks remained about the same as during the 90 minutes prior to the suspension. Over the same interval, mean relative quoted spread for large cap Nasdaq-listed stocks rose from 5.63 bp to 6.10 bp (an 8% increase). While also noisy, Nasdaq-listed stock spreads tended to decline for most of the remainder of the day.

Figure 11: Equal-weighted Mean Quoted Spread (%) for Large Cap Stocks on July 8, 2015

Taken together, Figures 9, 10 and 11, show that the increase in relative quoted spreads for NYSE-listed stocks was most pronounced for small cap stocks.

Impact on Depths

Figure 12 shows the mean equal-weighted inside depth in dollars for NYSE-listed stocks (red line), Nasdaq-listed stocks (orange line) and the NYSE/Nasdaq ratio.[7] Inside depth declined for NYSE-listed stocks a few minutes prior to 11:32 am and generally remained depressed until just before 3:10 pm. Mean inside depth fell from $87,659 to $69,712 (a 20% decline) and reached a minimum of $60,469 at 11:38 am. Depth for NYSE-listed stocks improved dramatically after the resumption of trading. Nasdaq-listed stocks showed a smaller decline immediately before and during much of the event window. Mean inside depth in Nasdaq-listed stocks fell from $38,368 to $37,120 (a 3% decline) and reached a minimum of $34,645 at 11:38 am. The difference between these two sets of stocks is evident in the relative change in depths, the NYSE/NDQ ratio (blue line).

Figure 12: Equal-weighted Mean Inside Depth ($) for All Stocks on July 8, 2015

Changes in the NYSE/NDQ ratio are also apparent in Figures 13 and 14, which chart average inside depths weighted by market cap and trading volume. However, these weighting choices generate similar mean inside dollar depth for NYSE-listed and Nasdaq-listed stocks, suggesting that smaller stocks exerted more influence on the overall level of the NYSE/NDQ ratio than on the time series behavior of the ratio.

Figure 13: Market Cap-weighted Mean Inside Depth ($) for All Stocks on July 8, 2015

Figure 14: Dollar Volume-weighted Mean Inside Depth ($) for All Stocks on July 8, 2015

Figures 15, 16 and 17 show mean equal-weighted inside dollar depth for stocks broken down by market capitalization. The same three size bins used in the quoted spread analyses are created from the 30^th and 70^th NYSE-listed market capitalization percentiles, and these cut offs are used to assign both NYSE-listed and Nasdaq-listed stocks to one of the bins.

Figure 15 shows inside depth for NYSE-listed and Nasdaq-listed small cap stocks was nearly the same at 11:32 am as it was at 11:05 am. NYSE-listed inside depth increased throughout the day. Inside depth for Nasdaq-listed small cap stocks remained fairly constant until after 2:00 pm and increased sharply just before the close.

Figure 15: Equal-weighted Mean Inside Depth ($) for Small Cap Stocks

Figure 16 shows inside depth for both NYSE-listed and Nasdaq-listed mid cap stocks declined in the minutes leading up to the trading suspension. However, by 12:30 pm depth recovered and tended to remain at or near those levels until the resumption of trading at 3:10 pm.

Figure 16: Equal-weighted Mean Inside Depth ($) for Mid Cap Stocks

Figure 17 shows that inside depth for NYSE-listed large cap stocks declined just prior to 11:32 am and remained lower throughout the trading suspension. There was a brief decline in depth for Nasdaq-listed large cap stocks just before 11:32 am. However, inside depth for Nasdaq-listed large cap stocks tended to rise throughout the remainder of the day.

Figure 17: Equal-weighted Mean Inside Depth ($) for Large Cap Stocks

Taken together, Figures 15, 16 and 17 show the decline in depths for NYSE-listed stocks was most pronounced for large cap stocks.

Liquidity Provision on the NYSE

Figure 18 shows the equal-weighted percentage of time that each exchange spent at the inside spread for each day during the period from January to July 2015.[8] NYSE, Nasdaq and Arca all quoted at the inside spread for NYSE-listed stocks over half of the day. On average NYSE spent over 70% of the day at the inside quotes, Nasdaq spent over 60% of the day at the inside and Arca spent over 50% of the day at the inside. The remaining exchanges (gray lines) all spent less than half of the day at the inside quotes for NYSE-listed stocks, on average. Overall, Figure 18 shows that while NYSE provided liquidity at the best quotes for NYSE-listed stocks most of the day, they were not the only exchange to do so.

Figure 18: Percentage of Time at the Inside Spread for NYSE-listed Stocks – January to July 2015

Figure 19 shows that, on average, NYSE was the sole provider of liquidity at the inside quotes between 7% and 11% of the day (between 27 and 43 minutes) for NYSE-listed stocks during the period from January to July 2015. Nasdaq and Arca were the sole providers at the inside between 2.5% and 7% of the day, and the remaining exchanges were solely at the inside for less than 2.5% of the day. On average, NYSE-listed stocks had only one exchange at the inside around 22% of the day (the sum of all of the lines in Figure 19), and, NYSE was the sole exchange between one-third and one-half of that time.

Figure 19: Time Alone at the Inside Spread for NYSE-listed Stocks – January to July 2015

Taken together, Figures 18 and 19 suggest that NYSE played an important role in liquidity provision for NYSE-listed stocks in 2015. For some of the day, NYSE was the sole provider of liquidity at the inside quotes for the average stock. For most of the rest of the day, NYSE provided liquidity at the inside along with other exchanges.

Figure 20 shows the mean percentage of the day that NYSE spent alone at the inside quotes for NYSE-listed stocks broken down into deciles by the share volume traded for each day. NYSE was alone at the inside between 10% and 25% of the day for stocks in the first through third least-frequently traded deciles (Share Volume Deciles 1 – 3) and was alone at the inside less than 3% of the day for the most-frequently traded decile (Share Volume Decile 10 – gray line).[9] NYSE’s importance in liquidity provision, measured by the amount of time NYSE is the sole liquidity provider, is more evident for less actively traded stocks.

Figure 20: Time Alone at the Inside on NYSE – January to July 2015 By Share Volume Deciles

Putting It Together

Figures 1 and 2 show that while trading volume in NYSE-listed corporate stocks might have been higher throughout the day on July 8 had there been no trading suspension, it was well within the daily range for 2015.

Figures 3 – 5 show that trading in NYSE-listed corporate stocks shifted to the remaining market centers, especially other exchanges. In short, it appears that traders migrated to other exchanges when the NYSE suspended trading on July 8.

Figures 6 – 8 show that, on average, relative quoted spreads for NYSE-listed stocks increased immediately prior to the suspension of trading and remained elevated for the next 75 minutes. When compared to mean spreads for Nasdaq-listed stocks, mean NYSE-listed spreads remained higher throughout the event window and declined (relative to Nasdaq-listed spreads) only after trading resumed in the afternoon. Figures 9 – 11 show that while this differential behavior between NYSE-listed and Nasdaq-listed stocks was evident in small, mid and large capitalization stocks, the starkest contrast was in the smallest stocks, suggesting that the trading suspension had the greatest impact on smaller and less actively traded NYSE stock spreads.

Figures 12 – 14 show that, on average, inside dollar depth declined for NYSE-listed stocks a few minutes before the suspension of trading while the impact on Nasdaq-listed stocks was less noticeable. The sharp increase in depth for NYSE-listed stocks (both nominal and relative to Nasdaq-listed stocks) after trading resumed in the afternoon is also evident in most charts.

Figures 15 – 17 show that large cap stocks were most responsible for the declining average depth of NYSE-listed stocks. Mid cap stocks did not show declining depth throughout the event window for either NYSE- or Nasdaq-listed stocks, and small cap stocks show NYSE-listed depth increased for most of the event window. In contrast to the spread results, the trading suspension did not exert a negative impact on quoted depth for smaller and less actively traded NYSE stocks.

Figures 18 – 20 show that, on average, NYSE spent about 70% of the day at the inside for NYSE-listed stocks during the period from January to July 2015, more than Nasdaq (60%) and Arca (50%). On average, NYSE spent between 7% and 11% of the day alone at the inside quote, again, more than any other exchange. NYSE spent over 10% of the day – and as much as 20% to 25% of the day – alone at the inside for the 30% least actively traded NYSE-listed stocks. NYSE was alone at the inside more frequently for stocks that did not trade as much.

During the period from January to July 2015, NYSE provided more sole liquidity at the inside quotes to less actively traded and smaller stocks. On July 8, these stocks experienced the greatest increase in spreads (on a relative basis) but did not experience a decrease in inside depth during the trading suspension. For larger and more actively traded stocks (the stocks for which NYSE provided the least sole liquidity), spreads widened a little and inside depth decreased a lot. NYSE’s absence reduced quoted inside depth but had less of an impact on spreads because presence at the inside quotes was maintained at other exchanges.

For smaller and less actively traded stocks (the stocks for which NYSE provided the most sole liquidity at the inside quotes during the period from January to July 2015), spreads widened but inside depth did not decrease on July 8. In fact, inside depth increased during most of the trading suspension. In other words, other exchanges not only quoted the same inside depth as before the NYSE absence, but also made up for the absence of NYSE depth. This depth, however, was provided at wider average spreads than typically available when NYSE is functioning normally.