Institutional Trading Costs

1
Institutional Trading Costs

• Rotman School Distinguished Lecture Series
• March 2008
• Ingrid M. Werner
• Martin and Andrew Murrer Professor of Finance
• Fisher College of Business, The Ohio State
  University

2
Introduction

• More than 70 of trading in developed markets is
  done by institutions.
• A block trade is defined (somewhat arbitrarily)
  as trades gt 10K shares.
• Block orders (institutional orders) are in
  practice significantly larger.
• Institutional traders need liquidity!!!
• The quote size is typically not sufficient to
  satisfy their trading needs.
• Displaying large order will result in adverse
  price impact.
• How do you tease out non-displayed liquidity
  without incurring price impact?

3
A Definition of Liquidity

• A market is liquid if
• uninformed traders can (Who)
• quickly (Immediacy)
• buy or sell (Symmetry)
• large size (Depth)
• when they want (Availability)
• at low transaction cost (Spread)

4
Other Variables Affecting Liquidity

• Security characteristics
• Recent trading history
• Resiliency
• The trader
• Informed or uninformed
• The market structure
• Competitive or oligopolistic dealers
• Order- or quote-driven

5
Liquidity is Valuable

• Liquidity helps traders solve cash flow problems.
• People and firms will save more and thereby
  decrease corporate costs of capital.
• Liquidity facilitates hedging strategies.
• More risks will be shifted between natural
  hedgers and production will be more efficient.
• Liquidity facilitates risk sharing schemes.
• More extremely large risky positive NPV projects
  will be undertaken.
• Positive externalities

6
Measuring Liquidity

• Quoted spread
• Effective spread
• Price impact
• Depth
• Trading volume

7
The Search for Liquidity

• Liquidity is the object of a bilateral search.
• Buyers search for sellers and sellers search for
  buyers.
• EBay is perhaps the greatest example of how an
  exchange/market can lower cost of search, and
  hence increase liquidity for sometimes obscure
  items.
• Exchanges can be thought of as search engines.
• Trader objectives affect their search strategies.
• Impatient traders demand liquidity.
• Passive traders supply liquidity.

8
Who supplies Liquidity?

• Patient pre-committed traders
• Buy-side institutions
• Individual traders
• Value-motivated traders
• Individuals
• Institutions
• Arbitrageurs
• Pure arbitrageurs
• Index enhancement funds
• Pairs traders

• Market makers
• Exchange specialists
• NASD dealers
• Scalpers
• Day traders
• Upstairs dealers
• Block positioners
• Package and program traders

9
Who uses Liquidity?

• Institutional traders
• Retail traders
• Professional traders

10
Institutional Traders and Liquidity

• The largest challenge facing institutional
  traders (buy-side traders) is finding liquidity.
• Quotes/order book depth is typically insufficient
  for large traders.
• Large orders tend to be perceived as informed gt
  displaying large orders may be costly.
• Recent market developments have contributed to
  diminishing displayed liquidity.
• Market fragmentation provides additional
  challenges (as well as opportunities) for
  institutional traders.

11
Types of Liquidity
Price

• Displayed
• NYSE book, ECNs
• Dealer quotes

• Not-displayed at market venue
• NYSE floor brokers
• ECN reserve orders

• Not-displayed at broker-dealers
• Agency orders held by brokers
• Not-displayed dealer liquidity

Most liquidity is not displayed

• Not-displayed at investors
• Orders at buy-side desks
• Latent liquidity

Source George Sofianos, Goldman Sachs
Quantity1
12
Dealing with Non-Displayed Liquidity

• Downstairs trading
• Split orders over time
• Split orders across dealers
• Upstairs trading
• Upstairs broker/dealer screens traders based on
  information
• Reputation secures the integrity
• Repeated interaction
• Who owns the stock?
• Bloombergs
• Thomson Financial
• Negotiation systems
• Liquidnet
• Liquidity quote (Nasdaq)
• Institutional Express

13
Block trades

• A block trade is any trade so large that it
  cannot be done easily using standard trading
  procedures.
• Block trades are often negotiated upstairs
  between block initiators and block liquidity
  suppliers.
• Block dealers
• Block brokers

14
The block trading problem

• Latent demand problem
• Order exposure problem
• Price discrimination problem
• Asymmetric information problem
• How can you convince a counterpart that you are
  uninformed?

15
Price Reaction to Block Trades

• The empirical literature has discovered that
  block trades move prices.

Temporary
Block buy
Permanent
Permanent
Block sell
Temporary
16
Easley and OHara (1987)Block Trades

• Why would informed traders use block trades?
• Model where trade size is chosen strategically by
  the informed traders.
• To make it simple, final value is either high or
  low and trade size is either large or small (not
  continuous).
• Two equilibria arise
• Semi-separating
• Zero small trade spread, positive large trade
  spread
• Pooling
• Positive small trade spread, positive large trade
  spread
• The semi-separating equilibrium arises only if
• The large trade size is large enough
• There is a low proportion of informed traders

17
Easley and OHara (1987)Block Trades

• How do market participants revise their beliefs
  based on order flow?
• Separating equilibrium (Bayesian updating)

No event uncertainty Small sell, block sell,
small trade
Add event uncertainty Small sell, block sell,
small trade
Matches evidence on price-reaction to block trade
in the literature
18
Saar (2001)Asymmetric Reaction to Blocks

• Why are buys associated with significantly larger
  permanent price impact than sells?
• Saar (2001) argues that three factors contribute
  to this pattern
• Mutual funds are reluctant to short sell
• Mutual funds cannot borrow to invest
• Mutual funds do stock research and they try to
  diversify
• Combined, these factors mean that there will be
  more information in stock purchases than sells.
• Specifically, if stock prices have increased (on
  average this is true), then mutual funds need to
  diversify and they pick stocks based on research
  gt
• The price impact for buys is larger than for sells

19
Burdett and OHara (1987)Building Blocks

• Models the search for a counterparties for a
  block trade.
• Suppose a institutional trader wants sell a large
  block
• Call a block dealer to negotiate a deal
• Block dealer may put together a syndicate
• Other block dealers may start short-selling in
  anticipation of the trade
• Front-running gt price will be depressed
• The block dealer sets a commission and a
  commitment price, then searches for a better
  deal.
• How long should the block dealer search?
• The longer, the higher the cost of front-running
• The shorter, the more likely you get a bad deal
• Model derives the optimal stopping time, at which
  point the block dealer buys the rest on his own
  account
• Has to sell it in the market at a discount
  (convex in size)

20
Burdett and OHara (1987)Building Blocks

• The commitment price is the expected value of the
  stock based on all publicly available
  information, which is also equal to the price
  posted by the specialist.
• The syndication has to result in a higher sell
  price for the institutional trader for it to add
  value.
• Why would buyers agree to pay more than the
  expected value of the stock?
• Overly optimistic traders
• Specialist price may be low because of inventory
  concerns (he may be long the stock) and the
  syndicate may be more diversified
• Perhaps traders are willing to pay up to do the
  block instead of a sequence of small trades
  (order processing costs)
• Realistic?

21
Grossman (1992)Upstairs versus Downstairs

• Can the upstairs market dominate?
• Upstairs market is specialized in seeking out
  unexpressed demand (non-displayed liquidity).
• Assume that the upstairs dealer has superior
  information on the state of the market.
• But, upstairs trading involves search costs and
  the delay is associated with a loss of utility.
• The main result is that if sufficiently many
  traders submit orders upstairs, the upstairs
  dealer can get an information advantage about
  displayed order flow gt the upstairs market will
  dominate

22
Seppi (1990)Order Splitting?

• Why dont block traders simply split their orders
  over time?
• Suppose that there is a single risk-neutral large
  trader that may be trading because of liquidity
  or information.
• There are uninformed traders, several competitive
  specialists, and several competitive block
  dealers.
• Important assumption
• Block trading is not anonymous (no bagging the
  street)
• Route informed order to block dealer gt penalty
  on subsequent trades gt compensates for losses on
  the (informed) block trade
• How would the large trader trade?
• Use block dealers when trading for liquidity
  reasons (b)
• Use specialists when trading for information
  reasons (xt, t1, 2, , T)
• Equilibria with randomization or informed trades
  also exist

23
Admati and Pfleiderer (1994)Sunshine Trading

• Traders who announce to the market who they are,
  what they intend to do, the full extent of their
  orders, and why they intend to trade.
• Would you believe the proclamations made by a
  sunshine trader?
• Are there potential adverse consequences of the
  order exposure implicit in sunshine trades?
• Model shows that this is optimal if a large
  trader wishes to trade for liquidity reasons in a
  market where there are also informed traders.

24
Madhavan and Cheng (1997)Upstairs versus
Downstairs Blocks

• Study gt20K blocks in Dow stocks 12/93-1/94.
• Compare the cost of upstairs versus downstairs
  trades, taking into account that traders
  optimally select where to send the order.
• Confirms prediction (Seppi (1990)) that upstairs
  trading is cheaper (but the difference is small)
  because liquidity traders can signal that they
  are informed.
• Estimate
• Permanent price impact (20,-1)
• Temporary Impact (0,20)
• Pre-trade leakage (-1, -20)
• Find that the permanent price impact of buys is
  much larger than for sells.

25
Madhavan and Cheng (1997)Upstairs versus
Downstairs Blocks
26
Madhavan and Cheng (1997)Upstairs versus
Downstairs Blocks
27
Madhavan and Cheng (1997)Upstairs versus
Downstairs Blocks
28
Bessembinder and Venkatamaran (2004)CLOB and
Upstairs Trading

• Paris Bourse has CLOB, why then would there be a
  need for upstairs trading?
• Most CLOBs actually do have a significant
  fraction of volume off-book.
• BV (2004) find support for Grossman (1990) that
  upstairs brokers tap into unexpressed liquidity.
• Costs for upstairs trades are much lower!
• Also find support for Seppi (1990) than upstairs
  brokers certifu trades as uninformed.
• Evidence suggests that upstairs market helps
  create liquidity
• Enhances market quality

29
The Importance of TCA

• A poorly executed trade can ruin the return on
  even the best ideas.
• Trading costs can be very high, especially on
  large orders in small cap stocks.
• Almost 2 percentage points on average!
• Plexus Group estimates that ¾ of research and
  portfolio management effort is lost in the
  transaction process.
• The easiest route to the top quartile of
  performance is to be in the bottom quartile of
  expense. Jack Bogle
• Trading cost analysis is a crucial part of the
  investment process.

30
What is TCA?

• The use of past order execution data to
• Analyze the factors influencing trading costs
• Evaluate execution quality
• Compare the costs of alternative strategies and
  execution venues
• Estimate expected trading costs
• Develop order execution strategies to minimize
  trading costs/maximize net returns (given a
  trading decision).

31
Why is TCA Important?

• The end of the bull market
• Lower returns
• Trading costs are more important
• Increased complexity of execution strategies
• Large orders
• Transition trades
• Proliferation of execution venues
• Where to execute?
• More choice more slippage?
• Regulatory pressure
• SECs regulatory agenda
• SEC Rules 11Ac-5 and 6 November 2000
• AIMR Trade Management Guidelines November 2002.

32
Small and Large Orders

• Trading costs for small orders
• Easy to measure
• Easy to interpret
• Easy to predict
• Main measures
• Commissions
• Quoted spreads
• Effective spreads
• Trading costs for large orders
• Difficult to measure
• Difficult to interpret
• Difficult to predict
• Main measures
• Commissions
• Implementation shortfall
• Deviations from VWAP

33
Direct and Indirect Trading Costs

• Direct costs
• Commissions
• Easy to measure
• Indirect costs
• Spreads, implementation shortfall, deviations
  from VWAP
• Hard to measure, especially for large orders

Indirect costs
Direct costs
Indirect costs
Direct costs
Large orders
Small orders
34
Quoted Spread

• MyFund sends a buy order for 1,000 shares of ABC
  stock
• Best offer price
• Lowest displayed price at which we can buy
• 25.42
• Best bid price
• Highest displayed price at which we can sell
• 25.38
• Quoted spread
• Best offer best bid
• 0.04 or 16 bps
• Roundtrip cost when we buy at offer and sell at
  bid
• Reward for liquidity providers
• Quoted depth
• 2,000 shares at the offer
• 2,000 shares at the bid

Offer
Quotedspread
Bid
35
Effective Spread

• Because there may be non-displayed liquidity
  inside the quote, small orders may execute at a
  price better than the quote
• Price improvement
• 0.01
• Definition of effective spread
• Buy orders (execution price minus mid-quote)2
• 0.02 or 8 bps
• Roundtrip costs if we trade taking into account
  the opportunity of price improvement

Execution price
Mid-quote
36
SEC Rule 11Ac-5 estimates(January 2005)
Universe SP 500 Component stocks Held
orders Less than 10,000 shares
Assuming a 30.00 stock
1 cent
0.033
Note NYSE argues that better matching shows
that the NYSE is cheaper
Source http//www.nasdaq.com/newsroom/stats/Perfo
rmance_Report.stm Note As of January 2005.
Source Market Systems, Inc. SEC 11Ac1-5 data.
Data are for all marketable orders, all sizes
under 10,000 shares, for SP 500 component stocks.
37
Large Orders

• MyFund issues a buy order for 600,000 shares of
  ABC stock
• Walk up the book using displayed liquidity
• Buy 2,000 shares at 25.42
• Buy 1,500 shares at 25.43
• Buy 1,000 shares at 25.44
• Buy 1,200 shares at 25.45
• Etc.
• Expensive
• Liquidity impact
• Access non-displayed liquidity and work order
  over time
• shop the order
• Send to upstairs broker-dealer to work over the
  day

38
Benchmarks for TCA

• The industry uses several alternative benchmarks
• Volume-Weighted-Average-Price (VWAP) of all
  trades over the period when the order was being
  filled (Abel/Noser).
• Closing prices (SEI).
• Implementation shortfall (Plexus Group).
• Calculate the value of a paper-portfolio based on
  having bought (or sold) the shares at the
  prevailing price when the decision was made.
• Usually, the benchmark price used is the
  mid-quote at the time the decision was made.
• Compare the value of the paper portfolio to the
  actual value of the portfolio after all orders
  have been filled (transaction costs).
• Multiply the unfilled sized by the difference
  between the current price and the benchmark price
  (missed trade opportunity cost)
• Average of the daily opening, high, low, and
  closing prices (Elkins/McSherry).

39
Profile of Large Buy Order
1310 execution completed
1230 second execution
1050 first execution
25.85
25.80
1045 order arrives at broker-dealer
25.75
25.70
945 MyFund decides to buy 600,000 shares of ABC
25.75 average execution price
25.60
25.40
Broker-dealer splits order in three 200,000-share
executions
25.35
Trader shops order
Source Goldman Sachs, Trading and Market
Structure Analysis
40
Implementation Shortfall
Average execution price
1045 order arrives at broker-dealer
25.80
945 MyFund decides to buy
25.75
ES
IS
25.60
PC
25.40
IS Implementation shortfall (including fees)
PC Pre-trade cost
ES Execution shortfall of Venue B
IS PC ES commissions
Source Goldman Sachs, Trading and Market
Structure Analysis
41
Factors affecting Implementation Shortfall
Liquidity impact Strain on liquidity caused by
the three executions
25.80
Information leakage Accelerates adverse price
move against MyFund
Fundamental-price move Change in price of stock
over execution horizon had MyFund not decided to
trade Can be favorable or unfavorable Not caused
by MyFund or broker-dealer MyFund faces an
unfavorable base-price move Fundamental-price
changes create execution risk
25.40
25.35
Source Goldman Sachs, Trading and Market
Structure Analysis
42
Chan and Lakonishok (1997)Trading Costs for
Institutional Orders

• Institutional orders are often broken up into a
  sequence of trades.
• To estimate the trading costs, we need
  information on the timing and size of orders.
• Plexus Group provided data on large trades for 33
  investment management (IM) firms, 1989-1991.
• Styles and strategies
• Buy and sell
• IM identities, broker IDs
• Commissions and Price impact
• Packages (5 days)
• Impact is measured compared to two benchmarks
• Pre open on the first day
• Post close 5 days after order is completed
• Remove market-wide movements using a
  size-controlled benchmark portfolio.

43
Order characteristics

• Stock liquidity
• Average daily share-volume
• Turnover
• Complexity/Difficulty
• Order size relative to normal trading volume.
• Dollar order size
• Urgency
• Risk increases with time to execution
• Risk aversion
• More volatility gt Increased urgency

44
Chan and Lakonishok (1997)Trading Costs for
Institutional Orders

• Objective is to compare institutional trading
  costs on Nasdaq to institutional trading costs on
  NYSE.
• Nasdaq orders are more difficult (complexity)
  than NYSE orders.
• More easy orders in NYSE stocks
• Sorting trading costs by characteristics gt less
  clear that costs on the NYSE are lower than those
  on Nasdaq gt more controls?
• Controlling for firm size, order complexity, IM
  identity, price, trading volume, etc, in a
  regression framework.
• Costs for trading small firms are lower on Nasdaq
• Costs for trading large firms are lower on the
  NYSE.
• Evidence that the price impact of buys is larger
  than that of sells is not clear

45
Chan and Lakonishok (1997)Trading Costs for
Institutional Orders
46
Chan and Lakonishok (1997)Trading Costs for
Institutional Orders
47
Chiyachantana, Jain, Jiang, and Wood
(2004)International Institutional Trading Costs

• Study a large cross-section of 37 countries,
  1997-1998 and 2001.
• Institutional order data from Plexus
• Document international institutional trading
  costs and the importance of benchmarks
• Confirm the prediction (Saar (2001)) that the
  price impact of buys (sells) exceeds that of
  sells (buys) after market run-ups (declines).

48
Chiyachantana, Jain, Jiang, and Wood
(2004)International Institutional Trading Costs
49
Chiyachantana, Jain, Jiang, and Wood
(2004)International Institutional Trading Costs
50
Liquidity impact, stock liquidity, and order
complexity
Closing price on day preceding the day the
decision to trade was made. Price prevailing
when institution releases order to trading desk.
Source Chiyachantana, C., P. Jain, C. Jiang, and
R. Wood, Institutional Trading Behavior and
Price Impact, Journal of Finance, 2004.
51
Trading costs for institutional-size orders (2001)
Emerging Markets
Developed Markets
Source Chiyachantana, C., P. Jain, C. Jiang, and
R. Wood, Institutional Trading Behavior and
Price Impact, Journal of Finance, 2004.
52
Chiyachantana, Jain, Jiang, and Wood
(2004)International Institutional Trading Costs
53
The Plexus Iceberg of Transaction Costs (US, 2007)

Direct Costs

Execution Shortfall
Pre-trade Costs (PC)
54
Managing the Process

• Lessons from the Plexus Iceberg
• Reducing execution shortfall and commission will
  not suffice to improve performance significantly.
• The firm also needs to work on delays (Pre-Trade
  Costs) and missed opportunities.
• They can be identified with proper record-keeping
• They can be reduced significantly
• Improving the process by shortening the pre-trade
  lifecycle of an order can enhance performance
  significantly.
• Monitoring incomplete fills, and providing proper
  order instructions to executing brokers will also
  significantly reduce trading costs.

55
A Solution

• Start with the existing portfolio
• Estimate the difference in performance between
  the desired purchases and the sales gt
  incremental performance
• Project the total costs (TCA) of buying and
  selling.
• If the estimated incremental performance exceeds
  the projected TC, complete the executions.
• If it does not, go back to the drawing board.
• Substitute high t-cost stock for highly
  correlated low t-cost stock?
• Track incremental performance

56
Conclusions

• The regulatory and market environment gt TCA is
  necessary
• Evaluating TC is relatively straight-forward (IS
  vs. VWAP)
• Predicting TC is challenging
• Order characteristics and order strategies go
  hand in hand
• Liquidity
• Complexity
• Urgency
• Work on the process to reduce pre-trade costs and
  missed trades
• Integrate the TCA into the portfolio allocation
  process

57
References

• Admati, A., and P. Pfleiderer, 1991, Sunshine
  trading and financial market equilibrium, Review
  of Financial Studies 4, 443-481.
• Bessembinder, H., and K. Venkatamaran, 2004, Does
  an electronic stock exchange need an upstairs
  market, Journal of Financial Economics 73, 3-36.
• Burdett, K., and M. OHara, 1987, Building
  blocks An introduction to block trading, Journal
  of Banking and Finance 11, 193-212.
• Chan, L., and J. Lakonishok, 1997, Institutional
  equity trading costs NYSE versus Nasdaq, Journal
  of Finance 52, 713-735.
• Chiyachantana, C., P. Jain, C. Jiang, and R.
  Wood, 2004, Institutional trading behavior and
  price impact, Journal of Finance.
• Grossman, G., 1992, The informational role of
  upstairs and downstairs markets, Journal of
  Business 65, 509-529.
• Madhavan, A., and M. Cheng, 1997, In search of
  liquidity An analysis of upstairs and downstairs
  trades, Review of Financial Studies 10, 175-204.
• Saar, G., 2001, Price impact asymmetry of block
  trades An institutional trading explanation,
  Review of Financial Studies 14, 1153-1181.
• Seppi, D., 2000, Equilibrium block trading and
  asymmetric information, Journal of Finance 45,
  73-94.