Daily and Intraday Patterns in Spread and Depth for Limit Orders and Specialists

Introduction

Much research in market microstructure focuses on trading costs-particularly the bid-ask spread,1 while less attention is paid to the depth of a quote (the number of shares offered at the bid and ask prices). The purpose of this paper is to expand upon the previous findings of daily and intraday spreads and depth. We examine quoted depth on a daily and intraday basis and further divide our sample into quotes submitted by the NYSE specialist and the limit order book. Little research focuses on depth on a daily basis and an intraday basis. This paper adds to the literature by examining depth on a daily and intraday basis and analyzing the origination of the liquidity (quoted depth)-the specialist or the limit order book.

There are several studies that examine quoted depth (Lee, Mucklow, and Ready, 1993; Harris, 1994; and Dupont, 2000). Lee, Mucklow, and Ready study aggregate depth around earnings announcements. They find that wide spreads are accompanied by low depth and credit this finding to the specialist's behavior when faced with asymmetric information. Additionally, a study by Harris and Dupont (2000) shows that spreads and depth are used together to manage adverse selection and inventory risks. Both of these studies examine quoted depth and do not distinguish between depth provided by the specialist and limit order book.2

Several studies differentiate the specialist's quoted depth from that of the limit order book (Kavajecz, 1999; Goldstein and Kavajecz, 2000; and Kavajecz and Odders-White, 2001). These studies do not, however, examine these differences on a daily or intraday basis. Kavajecz (1999) examines changes in quoted depth of the specialist and limit order book around informational events and finds that both limit order traders and the specialist reduce depth around informational events. Goldstein and Kavajecz study differences in liquidity (depth) associated with a reduction in the tick size on the NYSE and find that liquidity is reduced with a change in tick size. Kavajecz and Odders-White analyze changes in the posted quotes and find that changes in orders from the limit order book have significant effects on the posted quotes.

We seek to extend current research by looking at quoted depth on a daily and intraday basis for both the specialist and the limit order book. Additionally, we examine quoted spreads and depth for a day-of-the-week effect.3 Chorida, Roll, and Subrahmanyam (2001) (CRS) find daily differences in spreads and depth, but provide no explanations or reasons for these findings. Chorida, Roll, and Subrahmanyam find significantly negative (positive) coefficients for Tuesday, Wednesday, and Thursdays when examining daily spread (depth) changes from 1988 to 1998. We, like Chorida, Roll, and Subrahmanyam, find lower spreads and higher depth on Tuesdays and add to their findings by showing that this finding seems to come mainly from the limit order book in the latter part of the trading day. Additionally, our results show that the specialist provides less liquidity (as measured by quoted depth) than do limit orders.

Data and Quote Identification

The data for this study come from the NYSE TORQ (Trades, Orders, Reports, and Quotes) database.4 This database consists of time-ordered transactions for 144 selected stocks traded on the NYSE from November 1990 through January 1991.5 There are 63 trading days in the sample period. We filter the database using several screens. We omit the quotes if either the bid price or ask price is equal to or less than zero or the spread is greater than $4 or less than zero. Additionally, we limit our sample to stocks that have at least one trade in each of the 63 days in the sample. Our final sample consists of 108 stocks.

Following the quote classification procedure by Chung, Van Ness, and Van Ness (1999), all bid (ask) quotes are classified into one of three categories, according to whether the quote reflects the trading interest of the specialist, limit-order traders, or both. To determine where the quote is coming from, we partition the depth of each quote into that provided by the specialist, the limit-order book, or a combination of the two.

If a bid (ask) quote has no corresponding limit orders, the quote is identified as a specialist quote (quote class S). If a bid (ask) quote has one or more corresponding limit orders, we compare the quoted depth at the bid (ask) with the depth of the matching limit order(s). If the limit orders are equal to the quoted depth, we identify it as a limit order quote (quote class L). If the limit order depth is less than the quoted depth, we identify this quote as a mixed quote, partially from the limit order book and partially from the specialist (we call this quote class M).

Because the origination of each ask and bid is identified as the specialist, limit order book, or both, we follow Chung, et al. (1999) in categorizing each quote as a specialist quote or a limit order quote. If a bid or ask quote belong to the quote class S and/or M, the quote is designated as a specialist quote. Limit order quotes are those that belong to the quote classes L and M.6

Spread Measures

We use several spread measures in our analyses. We define the quoted spread as the difference in the ask and bid prices (spread = ask - bid).

Spreads and Depth by Day of the Week

Chorida, Roll, and Subrahmanyam (2001) undertake the most extensive analysis of daily changes in liquidity and trading. They find strong day of the week effects, and Tuesdays tend to show an increase in liquidity (decreased spreads and increased depth). We seek to expand upon their findings by examining spreads and depth on a daily basis for both specialists and limit orders. Each quote in the data is identified as either a specialist quote or a limit order quote (defined previously). Table 1 (panel A through E) presents spread, percentage spread, standardized spread, percentage standardized spread, and quoted depth for the specialist and limit order book. These statistics are calculated for each day of the week. We find that spreads and standardized spreads for limit order quotes are significantly lower on Tuesdays than the other days of the week. The average spread (panel A of table 1) for limit order quotes is 0.1844 and on Tuesdays it is 0.1818. An F-test of whether the mean on Tuesday is equal to the average of the other weekday means, (F^sub Tue^), rejects equality at the 5 percent level for limit order quotes. (Equality is rejected for both the mean quoted spread and mean standardized spread.)

The findings that Tuesdays have lower spreads is consistent with the findings of Chorida, Roll, and Subrahmanyam (2001), who also find that spreads are lower on Tuesdays than other days of the week. We also show that the decrease in spreads on Tuesdays is primarily attributable to limit orders. We also find that Monday has a significantly larger spread for limit order quotes than do the other days of the week.8

Lee, Mucklow, and Ready (1993) suggest that the lower spreads imply larger depth. Given our findings of lower spreads on Tuesdays (for limit order quotes), we expect, and find that significantly greater depth for the limit order book on Tuesdays than on the other days of the week (average quoted depth of 29,626 and a quoted depth of 31,700 on Tuesdays). Neither the specialist's quoted depth nor the spread measure is significantly different on any of the days of the week.

lntraday Depth

Table 2 examines the intraday depth for each of the quote classes. The largest average depth occurs when both the specialist and limit order book determine the quote. Quote class (M,M) averages 46,293 shares combined at the ask and bid prices. The smallest quoted depth occurs when the specialist alone has both the bid and ask quotes (quote class (S,S) averages 7,400 shares). When the limit order book has both sides of the quote there is more quoted depth than when the specialist has both sides (quote class (L,L) averages 16,493 shares). Generally, the intraday patterns of depth (for all categories of specialist and limit order book quotes) exhibit inverted U-shaped patterns, which is opposite to the observed intraday patterns of spreads found by Mclnish and Wood (1992), Chan, Christie, and Schultz (1995), and Chung, Van Ness, and Van Ness (1999). The inverted U pattern is consistent with the relation of low spreads and high depth found by Lee, Mucklow, and Ready (1993)9 and the findings regarding the cumulative depth of the limit order book shown by Chung, Van Ness, and Van Ness (1999).10

Panel A of Table 3 lists the average depth for the entire day, while panel B shows the average intraday depth for the specialist and limit order book. We find that the limit order book has the largest average depth, 29,626 shares, while the specialist provides a depth of 24,262 shares. Biais, Million, and Spatt (1995) show that traders place limit orders when the spread is large and the limit order book is thin,11 while Chung, Van Ness, and Van Ness (1999) find that the spread narrows as the accumulated limit order book grows. Hence, we expect to find that quoted depth is largest for the limit order book in the middle of the day since Chung, et al. find that the accumulated limit order book is greatest in the middle of the day and declines toward the close of trading. Our results in Table 3 confirm this expectation. The specialist and limit order book all exhibit a rough intraday inverted U-shaped pattern in depth (this finding is depicted in Figure 1). The specialist appears to be providing more depth in the middle of the trading day at the same time that the limit order book is providing a substantial amount of liquidity. This finding is consistent with Bondarenko and Sung (2003), who present a model of specialist participation and limit order book depth.

Depth Differences for Specialist and Limit Orders

Table 4 examines differences in depth for specialists and the limit order book. We find that the average depth for the specialist is significantly smaller than that of the limit order book at the 5 percent level. The findings of Table 4, that the differences in average quoted depth of the limit order book are greater than that of the specialist, cause us to question whether these relationships are consistent throughout the trading day. Table 5 presents the intraday differences in depth.

Consistent relations emerge when examining the intraday differences. The limit order book maintains a larger quoted depth than the specialist throughout the trading day, and this difference is generally increasing throughout the day until the last hour of trading. This difference also shows an inverted U-shaped intraday pattern.

Day of the Week Depth for Specialist and Limit Orders

In Table 1 we show that depth on Tuesday is different from the other days of the week. We expand upon this finding by analyzing the intraday quoted depth for the specialist (Table 6). The F-statistics show no significant differences in the days of the week or intraday intervals for the specialist.

Table 7 expands on the Table 6 by looking at the intraday depth of the limit order book. The F-statistics for quoted depth from the limit order book show that the greater overall depth observed on Tuesday can be attributed to larger limit order depth in the latter part of the trading day.

Conclusions

We examine weekday and intraday relations of spreads and depth for both the specialist and limit order book. We find that the limit order book plays an important role in providing liquidity by posting orders with higher depth than the specialist.

On an intraday basis, we find that the average depth exhibits an inverted U-shaped pattern for both the specialist and limit order book. Our finding reinforce those of Lee, Mucklow, and Ready (1993) who find that wide spreads are accompanied by low depth-intraday spreads exhibit a U-shaped pattern; hence, intraday depth exhibits an inverted U-shaped pattern. The difference in quoted depth from the specialist and limit order book is significant, and this difference is significant on an intraday basis.

We also find that spreads are lower and depth is higher on Tuesdays. We attribute the higher quoted depth to the limit order book posting larger cumulative depth toward the end of the trading day on Tuesdays.