ASX Optimal Spreads and Tick Size on the ASX - James McCulloch

Document Sample
ASX Optimal Spreads and Tick Size on the ASX - James McCulloch Powered By Docstoc
					Optimal Spreads and Tick Size on the ASX
                        James McCulloch
                        September, 2003

    This short paper summarizes the arguments for reducing tick size
 on the Australian Stock Exchange. The discussion is a summary of
 a research thesis into an optimal tick schedule for Australian Stock
 Exchange written by the author [10]. The detailed research that ac-
 companied the development of the arguments to reduce tick size is not
 reproduced in this paper.

The Current Tick Size Schedule
The Australian Stock Exchange (ASX) currently has 3 different minimum
price steps (tick sizes) when trading stocks.

   • The tick size for stocks priced below $0.10 is $0.001.

   • The tick size for stocks priced from $0.10 to below $0.50 is $0.005.

   • The tick size for stock orders priced $0.50 and above is $0.01.

This tick size schedule has pronounced effects on trading strategy, market
quality and execution costs for stocks of different prices

    This is due to the different percentage spreads for different priced stocks.
As an example, the tick size of 1 cent for stocks priced 50 cents and above
is examined for different priced stocks. The minimum percentage step size
between buyers and sellers (relative spread size) of a 50 cent stock is 200
basis points. For a $50 stock the minimum spread is 2 basis points. Bro-
kerage (institutional) for the execution of stock on the ASX is estimated to
be approximately 20 basis points. Therefore, for a 50 cent stock, the cost of
executing a market order and crossing the 200 basis point spread is 10 times
estimated brokerage costs. Conversely, for a $50 stock the cost of execut-
ing a market order and crossing the 2 basis point spread is 1/10 estimated
brokerage costs.

Problems With The Current Tick Size Schedule
From the example above, it can be seen that for low priced stocks, the cur-
rent tick size schedule may result in excessively large minimum percentage
spreads. These excessive spreads produce the following undesirable market

   • Reduced Market Quality. Excessively wide spreads due to tick
     size in high turnover stocks encourages principal traders to simultane-
     ously queue bid and ask limit orders in order to profit from the wide
     spread. This dramatically increases limit order queue size and limit
     order time to execution, forcing ‘natural’ buyers and sellers to execute

     expensive market orders (unnecessarily cross the spread). Another un-
     desirable side-effect of excessive spreads in high turnover stocks is prin-
     cipal traders gaming the opening call auction in order to exploit these

   • High Transaction Costs. A direct result of the current tick size
     schedule is high percentage spreads for lower priced stocks.

   • Decreased Market Turnover. High transaction costs lead directly
     to decreased market turnover.

Reducing Relative Spreads
The solution to the problems listed above is to decrease tick size for low
priced stocks.

The Proposed Tick Size Schedule
Research [10] suggests that the following modified tick size schedule is optimal
for all stocks priced $0.10 and above:

   • The tick size for stocks priced below $2.00 is $0.001.

   • The tick size for stocks priced from $2.00 to below $10.00 is $0.005.

   • The tick size for stock orders priced $10.00 and above is $0.01.

    The proposed optimal tick size schedule is graphed in figure 1 with the
existing tick size schedule for comparison.

                                                                               Proposed Tick Size Regime
                                                                                                            Current Tick Size Regime
                                                                                                          Proposed Tick Size Regime
Half Tick Relative Spread (basis points log scale)



                                                       0.01             0.1                      1                        10
                                                                              Stock Price (dollars log scale)

                                                       Figure 1: The Proposed Optimal Tick Size Schedule on the ASX

Benefits Of The Proposed Tick Size Schedule
Reducing the tick size for stocks priced $0.10 to $10 will have the following
beneficial effects on the market:

   • Improved Market Quality. The adoption of the proposed tick size
     schedule will eliminate excessively wide minimum spreads and remove
     the incentive for principal traders exploit these excessive spreads. So
     principal traders will not be motivated to stack limit orders or game
     the market open.

   • Lower Transaction Costs. Reduced spreads means lower transaction

   • Increased Market Turnover. The reduction in market transaction
     costs because of lower spreads for lower priced stocks will increase mar-
     ket turnover for those stocks. It has been estimated [10] that if the
     proposed optimal tick size schedule is adopted, market turnover will
     permanently increase by 16% for stocks priced $0.10 to $10 and 6% for
     all stocks. This increased turnover represents a positive for brokers and
     a positive for the ASX.

Other Effects Of Tick Size Reduction
The reduction in tick size will also have the following effects. Note that these
effects are not detrimental to market quality (see the following section).

   • Reduced Limit Order Queue Size (Market Depth). Research
     estimates that limit order queue sizes will decrease by an average of
     36% for stocks priced $0.10 to $10. For low priced, high turnover
     stocks it is likely that the decrease in limit order queue size (market
     depth) will be greater than 36%. This means that average trade sizes
     for these stocks will be reduced.

   • Increased Multiple Tick-Size Bidding. For low turnover stocks
     the new tick sizes will be ‘too small’ and traders will compensate by
     quoting in multiple tick size increments. For example, a low turnover
     $1.00 stock may still be quoted in increments of 1 cent even though
     the new tick size will be 0.1 cent. Thus the stock is being quoted and
     traded in increments of 10 ticks. It is estimated that at least 69 stocks

         and probably over 100 stocks will be quoted and traded in multiple tick
         size increments as a result of the proposed tick size changes.

A Discussion Of Arguments Against Tick Size Reduc-
There are several arguments against reducing tick size that are considered
below and rejected.

       • Large Spreads Are Required To Motivate Principal Traders
         to Act as Liquidity Suppliers. The ASX already has high turnover
         stocks with low percentage spreads such as CBA, NAB and RIO. Hav-
         ing minimum spreads of less than 5 basis points does not seem to have
         harmed the liquidity of these very high turnover stocks.

       • The Proposed Tick Size Will Be Too Small For Low Turnover
         Stocks. This is true and these stocks will experience the ‘multiple
         tick-size bidding’ described above, but there is no evidence that ‘mul-
         tiple tick-size bidding’ is harmful to the market in any way. Note that
         the stocks that experience ‘multiple tick-size bidding’ will all be low
         turnover stocks and it is estimated that these stocks will be less than
         2% of market turnover.

       • The Reduced Spreads Will Result In Reduced Market Depth
         (Limit Order Queue Size). This is true and means that large orders
         cannot be executed immediately. However, this does not imply in a
         reduction in market quality.
         Low spread stocks such as CBA, NAB and RIO also have low market
         depth as a percentage daily turnover 1 but these stocks have very high
         turnover and liquidity. In fact on the ASX there is a positive correlation
         between higher stock stock price (lower relative spread and smaller
         market depth) and higher turnover. So lower spreads and lower market
         depth do not imply lower stock liquidity.
         This can be explained by the presence of ‘dynamic liquidity’ on the
         ASX. A dynamically liquid market is able to supply fresh limit order
    Typically these stocks have limit order queue sizes that average only about 0.3% of
daily turnover.

         inventory at the same or close to the same price within a short pe-
         riod of time after existing limit order inventory is exhausted. In other
         words, dynamically liquid markets have a self-regulating liquidity sup-
         ply process that is able to regain its prior state quickly after a liquidity
         shock. Large orders can be executed on dynamically liquid markets
         with minimal trading cost by the simple expedient of splitting the or-
         der into a number of smaller orders2 . Operators can use the dynamic
         liquidity of the ASX to ‘work’ the order across the day with the ad-
         vantage that transaction costs are lower because relative spreads are
         lower. This benefits everybody because lower transaction costs means
         higher turnover.

       • Reduced Spreads Will Trivialize Time Priority Limit order traders
         can gain priority in the limit order queue cheaply by simply ‘ticking
         up’ their orders by a trivially small price increment. This will act as a
         disincentive for limit order traders to display their orders in the limit
         order queue.
         Again this is not supported by the evidence when examining existing
         stocks with very low spreads such as CBA, NAB and RIO. Having small
         tick sizes means that more than one price level on the bid and ask side
         is ‘active’ and limit order traders can choose how ‘close’ to the market
         to enter their orders and trade-off execution price for expected time to
         execution (or vice versa). By contrast, large spreads mean that only
         one price is active on the bid and ask and queue time priority becomes
         important in determining expected time to execution.

The Evidence from Other Markets
There have been many examples over the past decade of stock exchanges
reducing minimum tick size. The most notable example was the New York
Stock Exchange (NYSE) which reduced tick size from $1/8 to $1/16 in June
1997 and then decimalized by decreasing tick size to 1 cent in Jan 2001. Other
exchanges to lower tick size include the Tokyo Stock Exchange (TOKYO),
the Toronto Stock Exchange (TSE), the American Stock Exchange (AMEX),
the Paris Bourse (PARIS) and NASDAQ.
    An additional motivation for operators to split up large institutional orders and trade
across the day is the VWAP quality of execution metric often used by institutions.

    These tick size reductions have created a minor industry in empirical
research papers as market microstructure researchers have used pre and post
tick size reduction market data to empirically examine the effects of tick size
reduction. These studies have examined the effect of the reduction in tick
size by examining the consequent reduction of relative spreads (both effective
and quoted) and changes in market microstructure with particular reference
to changes in limit order queue size and any changes in turnover or market
    Table 1 summarizes ten of these studies (the arrows in the table show the
direction of change with a reduction in tick size and the numbers in brackets
show the magnitude of change).
 Reseacher(s)                        Market          Spread         Depth       Turnover

 Ahn, Cao and Choe [2]             AMEX              ↓(16%)   No Effect          No Effect
 Ronen and Weaver [11]             AMEX              ↓(11%)   No Effect         ↑(liquidity)
 Goldstein and Kavajecz [8]         NYSE             ↓(14%)    ↓(48%)
 Jones and Lipson [9]               NYSE             ↓(19%)    ↓(28%)          ↑(qualified)
 Bollen and Whaley [6]              NYSE             ↓(21%)    ↓(37%)            ↑(12%)
 Bacidore [3]                        TSE             ↓(23%)    ↓(42%)           No Effect
 Ahn et al [1]                     TOKYO           ↓(24%-60%)     ↓             No Effect
 Bacidore et al [4]                 NYSE             ↓(24%)    ↓(50%)          ↑(liquidity)
 Bessembinder [5]                NYSE/Nasdaq         ↓(54%)    ↓(24%)
 Bourghelle and Declerck [7]        PARIS           No Effect      ↓             No Effect

 Proposed Tick Schedule              ASX             ↓(32%)        ↓(36%)       ↑(5.9%)

     Table 1: A Summary of Empirical Tick Size Reduction Research

    All the empirical tick size reduction studies, with the exception of Bourghelle
and Declerck [7], show that reduced tick size decreases relative spread and
market depth. These results are also consistent with the predicted decrease in
spreads and market depth (limit order queue size) on the ASX if the optimal
tick size schedule is adopted.
    Some researchers have claimed that because reduced tick size will re-
sult in reduced market depth, this will benefit smaller traders but lead to
higher transaction costs for larger traders. In the empirical research tabu-
lated above, the evidence that a decreased instantaneous market depth leads

to higher costs for larger traders is mixed and may not apply to an electronic
exchange like the ASX. Jones and Lipson [9] and Goldstein and Kavajecz [8]
both finding that the reduction of tick size from $1/8 to $1/16 on the NYSE
increased execution costs for large orders. Balanced against this Bacidore [3],
Ronen and Weaver [11], Bessembinder [5], Bacidore, Battalio and Jennings
[4] and Ahn, Cai, Chan and Hamao [1] all find that execution costs improved
or remained constant for large orders. The Bacidore [3] (TSE) and Ahn, Cai,
Chan and Hamao [1] (TOKYO) studies were both performed on electronic
exchanges closer in structure to the ASX than studies of Jones and Lipson
[9] and Goldstein and Kavajecz [8] which were on the open-outcry NYSE.
    Bourghelle and Declerck [7] study a modification of the tick size schedule
on the Paris Bourse in 1999. They find no subsequent reduction in spreads
and argue as a result that tick size reductions do not always result in an
increase in market quality. These results are important because the Paris
Bourse is an electronic market with a similar structure to the ASX. The
reason why Bourghelle and Declerck did not observe spread reductions be-
comes apparent if the new tick size schedule is compared to the old tick
size schedule. The new tick size schedule ‘was designed so that the relative
price increments did not exceed 0.10% of share prices’ and [the old tick size
schedule] ‘was only 0.05%, it reached 0.20% for stock priced just over 500FF’
(both page 3 [7]). Simply put, the relative tick sizes on the Paris Bourse were
already very low and from research on spreads [10], it is known that spreads
will only decrease if the minimum relative spread is larger than the ‘natural’
spread of the stock. With tick sizes of 5 basis points to 20 basis points in
the old tick size schedule, the Paris Bourse had already largely eliminated
excessive spreads due to tick size and therefore effect of further decreasing
tick size would have been small and difficult to observe. This also shows that
for electronic markets such as the ASX there is no benefit to be gained from
decreasing tick sizes further than is necessary ro remove excess spreads.

Implementation Of The Proposed Tick Schedule
Important Caveat
Markets such as the ASX have complex dynamics and it may be difficult to
model all the subtleties and interactions of market parameter interactions.
In particular, a change of a major market parameter such as the tick size
schedule may have consequences not identified in research and modelling.

If the ASX decides to proceed with reducing spreads, it is recommended
that this reduction be implemented cautiously. Perhaps initially with a trial
tick size reduction of a subset of high and low turnover stocks.

 [1] Hee-Joon Ahn, Jun Cai, Kalok Chan, and Yasushi Hamao, Tick Size
     Change and Liquidity Provision on the Tokyo Stock Exchange, Working
     Paper Archive at the University of Southern California, 2002.

 [2] Hee-Joon Anh, Charles Cao, and Hyuk Chao, Tick Size, Spread and
     Volume, Journal of Financial Intermediation 5 (1996), 2 – 22.

 [3] Jeffery Bacidore, The Impact of Decimalization on Market Quality: An
     Empirical Investigation of the Toronto Stock Exchange, Journal of Fi-
     nancial Intermediation 6 (1997), 92 – 120.

 [4] Jeffery Bacidore, Robert Battalio, and Robert Jennings, Order Submis-
     sion Strategies, Liquidity Supply and Trading in Pennies on the New
     York Stock Exchange, Working Paper. Available at the NYSE website,

 [5] Hendrik Bessembinder, Trade Execution Costs and Market Quality after
     Decimalization, Journal of Financial and Quantitative Analysis, forth-

 [6] Nicolas Bollen and Robert Whaley, Are ‘Teenies’ Better?, Journal of
     Portfolio Management 25 (1998), 10–26.

 [7] David Bourgelle and Fany Declerck, Why Markets should not Necessarily
     Reduce Tick Size (March 2002), Working Paper. Available from Fany
     Declerck’s working paper archive website, 2002.

 [8] Michael Goldstein and Kenneth Kavajecz, Eighths, Sixteenths and Mar-
     ket Depth: Changes in Tick Size and Liquidity Provision on the NYSE,
     Journal of Financial Economics 56 (2000), 125 – 149.

 [9] Charles Jones and Marc Lipson, Sixteenths: direct evidence on institu-
     tional execution costs, Working Paper, Available from the Journal of
     Financial Markets Website, 2000.

[10] James McCulloch, True Spreads and Optimal Tick Size on the Aus-
     tralian Stock Exchange, PhD Thesis, University of Technology Sydney,

[11] Tavy Ronen and Daniel Weaver, Teenies Anyone?, Journal of Financial
     Markets 4 (2001), 231–260.


Shared By:
Description: ASX Optimal Spreads and Tick Size on the ASX - James McCulloch ...