Market Structure
CEFA Module 1 Monday 3.9.2007 Mats Hansson
i. Aim
After this session you should (know):
Why market structures matter for finance professionals What kind of market structures exist In more detail how the two most important structures (dealer/market maker markets and LOB markets) work, their main features, their drawbacks and their advantages What “liquidity” is and how it can be measured What direct and indirect transaction costs exist How prices are formed (price discovery), and how market structure affects price changes (short term volatility) What is the role of “informed” and “uninformed” investors for liquidity and price discovery What is “transparency”, it’s importance, and if it’s good or bad The most important features of OMX Helsinki Be able to understand literature and research on market microstructure issues
ii. Scope and approach
“Market microstructure” is a broad topic:
An extensive body of academic literature exist on all subtopics listen in the “Aim” Many topics and details might have professional relevance/be interesting and would deserve more thorough treatment Unfortunately there is a time limit
The approach is to introduce you to the most relevant topics to get a “bird’s eye view” of the important issues in market structure instead on focusing on one or two After this you should know enough to ask the right questions and know how different issues are related The lectures start on a general/theoretical level to move to more practical/detailed treatment Focus on equity markets, but analysis valid on any market Limitations: this is the first time this material is lectured in Finnish
iii. Literature
McInish, T. H. (2000): “Capital Markets: A Global Perspective”, Chapters 2-3 Hansson, M (2007): “Market Structure”, CEFA Lecture Notes + Any additional material distributed
1. Introduction and Motivation
“Market microstructure”: the effect of financial market structure and trading rules on:
1. 2. 3. 4. 5. Transaction costs Liquidity Price discovery Volatility Transparency and information
Motivation 1: Financial economics/asset pricing
The usual assumptions behind financial models: 1. 2. 3. 4. No transaction costs/frictionless markets The ability to trade any amount (without affecting prices) No information asymmetries Market prices/returns that reflect true intrinsic value
are usually violated to some extent, depending on market structure
1. Introduction and Motivation
Motivation 2: Practical/useful
1. 2. 3. 4.
5.
What market structures exist? How is a trade executed on different markets? What transaction costs exist on different market structures? Are prices fair? Do they reflect true value, and can they be manipulated? Be able to discuss with and understand brokers and dealers and evaluate what they are saying
Motivation 3: For further analysis/discussion (?)
How well do different market structures handle different market situations? Why are different markets structured differently? Who do you trade with, and what is your counterparty’s motivation for trading with you?
2. Market structures: Classification
A market can be classified depending on:
1.
Who provides liquidity?
i) dealers (Quote driven markets) ii) the market (Order driven markets)
2.
Is trading continuous or not?
i) call markets: all buy/sell orders are executed at the same time, usually at one price which maximises trading volume ii) continuous markets: the market is open for a longer period of time, trading is possible at any time within trading hours, the price can change during the trading day (intra-day volatility)
Many markets are “hybrids” Also: trading floors (“crowd trading”, “pit trading”)
2. Market structures: Dealer markets
1.
2.
Dealers stand ready to by or sell by posting quotes:
the dealer buys into his/hers inventory at the bid-quote the dealer sells from his/hers own inventory at the ask quote
The dealer’s ask is always higher than the bid, for example Bid: 99.50, Ask: 100.00 The dealer makes his/hers income from the spread, i.e. ask – bid = spread, in this case: 100.00 – 99.50 = 0.50 , or as % from the midquote:
(100.00 + 99.50) / 2 = 99.75 (midquote)
0.50 / 99.75 = 0.005 = 0.5%
2. Market structures: Dealer markets
Examples of dealer-markets (automated or “manual”):
1.
2. 3.
4.
NASDAQ (automated, many market makers for each stock) Foreign exchange market Bond and money markets Derivative markets (interest rate and currency products)
Examples of hybrids with market makers:
1.
London Stock Exchange
SETS: LOB SETSmm: LOB with market making SEAQ: quote driven (market makers)
2.
NYSE
one “specialist” (market maker) each stock also has LOB
2. Market structures: Dealer markets
The role/obligations of the dealer/market makers differs across markets:
On some markets, there are no barriers to enter the market as a dealer and no obligations for the dealer. Examples are the foreign exchange, bond markets and (some) derivative markets. No obligations for the dealer might lead to the market “disappearing” or spreads becoming large when markets are volatile. On most stock markets operating at least partly as dealer markets dealers must become members of the exchange and pay fees to be allowed to trade as market makers. The market then usually imposes obligations on market makers, for example maximum spreads, minimum number of shares at the bid and ask, require quotes to be “firm”, require market making to be continuous and market makers to stay in the market for a certain period of time etc.
2. Market structures: Dealer markets
When the market is not an organised market (for example and exchange), financial information providers such as Bloomberg and Reuter’s collect and display dealer’s quotes:
Note that since the quotes of the dealers shown are not equal, we have both dealer spreads and a market spread
2. Market structures: Dealer markets
1.
SEB has the best bid with quotes 92.261 – 92.361 and has a spread of 0.10 or 0.108% Commerzbank and CS share the lowest ask (92.310), and hence the market spread is 92.261 – 92.310 or 0.049 Why are the quotes of the dealers not equal and what will happen in this market situation?
Quotes might differ for two reasons: i) dealers have different opinions about the value of the bond, ii) dealers are trying to affect orderflow to attract/try to avoid either buys or sells For example, we note that SEB quotes slightly higher than the two others. This might be because: i) SEB tries to attract sellers with a high bid, ii) SEB does not want to sell, and posts a high ask (investors will “shop around” for good prices). CS:s low bid might be because they have a larger than desired inventory of the bond and do not want to buy more. According to research, market makers are mostly active on one side only.
2.
3. Market structures: Limit order book markets
A limit order book (LOB) is a continuous electronic trading system where investors submit buy or sell orders with a certain price (hence the name “limit” order) and quantity All bid orders constitute the “bid side” and sell orders the “ask side”, making a LOB look like this (simplified view):
3. Market structures: Limit order book markets
Investors submit their orders through brokers who charge a commission for providing this service A broker is a member of the exchange with rights to trade and access the LOB and all trading information On a dealer market investor A trades with the dealer, while in a LOB, investor A trades with investor B and the broker(s) only act as an intermediary Hence, in a LOB liqudity is provided by the investors (the market), not by dealers/market makers Trading rules might also allow brokers to:
1.
2.
Trade for their own account for trading profit Enter liquidity providing (LP) contracts with issuers, and hence act as as market makers in the LOB
3. Market structures: Limit order book markets
Orders are usually executed in:
1.
2.
Price priority. The best price executes first, e.g. the best (highest) bid order is executed before lower bids. Time priority. If there are several orders with the same price, the one submitted first gets executed first.
All submitted orders are “firm” and are immediately executed at any time when a matching order arrives Orders can be withdrawn from the LOB If a matching order(s) already exists in the LOB, the submitted order is immediately executed against this order For example, if a bid at 40.84 of 80 shares is submitted, this order is immediately executed against the existing sell order at 40.84 of 80 shares
3. Market structures: Limit order book markets
Trading rules usually allow an order to be:
1.
2.
3.
Partly filled. E.g. an order of 200 shares is partly executed against the existing 80 shares, and the rest (120 shares) is either transferred to the bid side at the price 40.84 (the best bid –ask has then changed to 40.84 – 40.90), or waits with the broker for further action. Filled against several orders at the same price level. Filled against several orders at different price levels if the price of the order allows it. E.g. an order to buy 300 shares at 40.90 first clears the order at 40.84 of 80 shares and then against 220 shares of the sell at 40.90, leaving 280 shares left at 40.90 on the ask side. The best bid-ask has now widened to 40.70 – 40.90. The average price paid on these 300 shares are
80 x 40.84 + 220 x 40.90 = 12 265.20 / 300 = 40.884
4. Market structures: Hybrids/multiple systems
A market usually provides several ways of trading:
NYSE:
1. 2.
A “specialist” (market maker) for each stock, and A limit order book OMX Helsinki and many other LOB markets have:
1.
2.
A limit order book The possibility to trade “block trades” (large trades) outside the order book
Many bonds can be traded:
1. 2.
Over the counter (dealer market), or On an exchange
4. Market structures: Hybrids/multiple systems
Why? Since there is no optimal market structure that would suit all types of trades and investors Example: The Paris Bourse tried to force all trades into the LOB by not allowing block trades outside the LOB.
Consequence 1: large trades in dual-listed shares migrated to London’s SEAQ Consequence 2: block trading outside the LOB was again allowed
Questions arise:
Why is there not an “optimal” trading structure? Why do not large investors want to trade blocks in the LOB, and instead “negotiate” trades “upstairs” Why are not all trades (also small) “negotiated” trades?
4. Market structures: Hybrids/multiple systems
1.
For small, frequent trades, a LOB is usually efficient:
Execution is fast since the LOB usually provides sufficient liquidity to absorb the order fully without price impact Execution is cost-efficient since the order just has to be entered into the LOB, no counterparty search needed The investor submitting the order is unlikely to be worried about the “information content” of the trade once it is made visible to other traders in the LOB
2.
3.
1.
For large trades, a LOB might not work:
There is not enough liquidity in the LOB. Suppose an investor wants to buy 20 000 shares in Metso. There are only 12 935 shares (visible) on the ask side, the last is 500 shares at 41.57. A market bid of 20 000 shares will have a large price impact (about +2%) Even if there is some liquidity, it will take time to execute the order fully, and there will be uncertainty about the price.
2.
4. Market structures: Hybrids/multiple systems
3.
A large order has informational content. Large investors usually have more information than small investors. Hence, other investors might think “there is something going on”, get cautious, and withdraw sell orders or increase their sell prices. Trading might altogether cease, as the market absorbs this unusually large interest to buy shares.
1.
2.
Forcing blocks to the LOB might harm the whole market Solution: allow block trades to be negotiated upstairs:
Broker receives buy order from investor Broker contacts potential sellers among his/hers own clients, if not, using his/hers contacts among other brokerage houses Price is negotiated such that both parties agree on the price The trade is made outside the LOB, but must be reported as a transaction within a specified amount of time (for example 5 minutes), to provide information to other investors (fairness)
3.
4.
4. Market structures: Hybrids/multiple systems
1.
Further on why there are differences in microstructures:
A LOB might not work if trades are small but infrequent: difficulties in matching buys with sells, large spreads. Here, a market maker might increase liquidity. A LOB might not work if trades are large and infrequent. An example is the bond market (trades usually in millions of USD or EUR), where the only participants are large investors. Hence, i) there is no need to take small investor’s needs into consideration, ii) transactions likely to be made faster through broker’s contacts (dealer market), than through a LOB. There might be a resistance to change to more modern trading systems by those who make a living of knowing and using the old system.
2.
3.
5. Market structures: Call markets
Call markets are typically used when there is uncertainty/little information about the true value of the stock (fundamentals):
1.
2.
3.
To open continuous markets: it is easier to get continuous trading going once a price level has been established in the opening call (nobody wants to place the first order in an empty LOB) To trade very illiquid stocks: trading is more efficient if all liquidity is concentrated to one point in time When there is not sufficient technology to arrange continuous trading (up to 1989 the Helsinki Stock Exchange was a call market)
Call markets also have the advantage of fairness, since everyone (informed and uninformed investors) gets the same price if a uniform call price is used
6. Liquidity and Trading costs
1.
2. 3.
“An asset is liquid if it can be converted to cash easily” More analytical measures by Harris (1990):
Width: the spread Depth: size of trade that can be traded at a price level Immediacy: speed of transaction at a given size and cost Resiliency: speed at which prices return to initial level after they change in response to a large trade
4.
1. 2.
Trading costs:
Direct costs: commissions, taxes, acquiring information etc. Indirect costs:
• •
• •
Spread: buying above true value, selling below Market impact: the market moves in adverse direction due to price impact of executing a (large) order Market timing costs: market moves in adverse direction before order reaches the market/can be executed Opportunity cost: potential profit lost for orders not executed
6. Liquidity and Trading costs
Example of liquidity:
Spread: 0.01 EUR or 0.05% (from midquote 21.045) Number of shares at best bid and ask: 19 081 and 44 279 EUR volume at best bid and ask: 401 464 EUR and 932 073 EUR
6. Liquidity and Trading costs
More measures of liquidity, using the bid side as example:
Orders 5 4 11 10 13 10 11 4 4 7 BID Shares 19 081 35 254 97 500 69 800 141 128 47 461 101 968 11 045 13 350 81 400 Price 21.04 21.03 21.02 21.01 21.00 20.99 20.98 20.97 20.96 20.95 Price ch. Average Average (Midq.) P P ch. 0.02 % 21.0400 0.02 % 0.07 % 21.0335 0.05 % 0.12 % 21.0248 0.10 % 0.17 % 21.0202 0.12 % 0.21 % 21.0123 0.16 % 0.26 % 21.0097 0.17 % 0.31 % 21.0038 0.20 % 0.36 % 21.0031 0.20 % 0.40 % 21.0020 0.20 % 0.45 % 20.9952 0.24 % EUR EUR Shares At level C umulative C umulative 401 464 401 464 19 081 741 392 1 142 856 54 335 2 049 450 3 192 306 151 835 1 466 498 4 658 804 221 635 2 963 688 7 622 492 362 763 996 206 8 618 698 410 224 2 139 289 10 757 987 512 192 231 614 10 989 601 523 237 279 816 11 269 417 536 587 1 705 330 12 974 747 617 987
For example, 512 192 shares can be sold at a value of EUR 10 757 987, at an average price of 21.0038 The average price is paid for these shares is about 0.04 EUR or 0.20% lower than the midquote
6. Liquidity and Trading costs
What determines the spread on a dealer market?
1.
2.
3.
Asymmetric information. The risk of making losses when trading with “informed” investors. Widening the spread decreases probability of buying too high and/or selling too low. Order-processing costs. Simply the cost of running a market making business: membership fees, employees, rent etc. Inventory costs. The cost of holding an undiversified portfolio of inventory.
Stoll (1989): the most important component is orderprocessing costs (contributes to 47% of the spread), then asymmetric information (43%) and inventory costs (10%) Inventory costs small because dealers can easily rebalance the portfolio after large buys/sells on inter-dealer markets
6. Liquidity and Trading costs
1.
What determines the spread on a LOB market? No inventory or order-processing costs in spread Asymmetric information costs now borne by investors There will always be a spread in a LOB since there is a “gravitational pull” (Cohen et al. 1981) towards the other side of the market Suppose the spread in a LOB is 40.70 – 40.84 and an investor wants to buy. Alternative strategies:
A bid below best bid (< 40.70): price is low, but probability of execution also low A bid at the best bid (40.70): price higher but probability of execution also higher (at least one bid gets executed before) A bid above best bid (> 40.70), below best ask (< 40.84): price higher, probability again higher A bid at 40.84: probability of execution is 100%
2.
3.
4.
6. Liquidity and Trading costs
1.
According to Cohen et al. 1981 strategy 3. is usually the worst one since:
Probability of execution is still uncertain, but price paid would still be high and close to best ask Only a small increase in bid price is required to hit the other side of the market (ask) and get immediate execution, hence the benefit of certain execution outweighs the higher price This prevents the spread from becoming infinitely small
2.
3.
According to Biais et al. (1995), when the spread is tight investors hit the other side of the book, when it is wide they place limit orders (Paris Bourse) On a dealer market, the dealer’s quotes provide “immediacy”
6. Liquidity and Trading costs
Empirically, spreads on both dealer and LOB markets are:
Positively correlated (increse with)
1. 2. 3. 4. Volatility: uncertainty about the true price, leads to non-execution costs and greater informational asymmetries Average trade size: measures informed trading, increases informational asymmetries Tick size in proportion to share price: If a share costs 1.00, and the minimum tick is 0.01, the spread is at least 1% Turnover (amount of shares traded of all outstanding shares): high turnover indicates informed trading
Negatively correlated (decrease with)
1. Trading volume: volume means competition for orders and decreased informational asymmetries (every order and trade conveys information) Number of trades: frequent trades also mean competition among investors in a LOB, for market makers frequent trades mean easier inventory management and frequent update of information
2.
7. Price discovery, volatility, and information
1.
2.
Which market provides the “best” price discovery? Difficult to measure empirically since:
The true price is unknown Price changes depend on: i) fundamentals, ii) investor behavior, iii) market structure: market maker/investor shortterm behavior, spread, liquidity, “bid-ask bounce” etc.
1.
What is the role of “informed” and “uninformed” investors?
“Informed” investors are investors who posess better information about the true value of the stock than others: i) their trading provides the market with new information, ii) on average, they trading is profitable (at the expense on others) “Uninformed” investor have less information, their trading might even “distort” the market and prices, hence the name “noise traders”. They are however needed since they provide liquidity to the market. If their losses become too large, they will eventually leave the market, hence the importance of fairness.
2.
7. Price discovery, volatility, and information
1.
Research is dominated by the role of the market maker for price discovery:
Market makers/dealers stabilise prices by i) giving quotes and ii) absorbing trades into inventory (as opposed to LOB markets which immediately clears orders) The quotes are on the other hand affected by inventory, the true price might not be halfway between bid and ask (e.g. Smidt 1971, Garman 1976) Do market makers have above average information? According to empirical evidence from NYSE and NASDAQ they do not (Madhavan 2000), inventory increase does not lead price increase. Market makers update their information and prices as they trade with informed traders. This learning process is slower if: i) the proportion of uninformed traders is large, ii) market depth and volume is high as it takes more time to move prices (Easley & O’Hara 1987)
2.
3.
4.
7. Price discovery, volatility, and information
5.
6.
7.
A single market maker (NYSE specialist) might experiment with prices (move prices and spreads) to learn from the informed traders response to different prices, making price discovery faster than in a multiple market maker market (Leach & Madhavan 1993) With multiple market makers, price experimentation is usually not feasible as there is competition and a market maker might not be able to make up for experimentation losses later in the session like a specialist (see quote above) Hasbrouck (1995) statistically decomposes the observed return process (NYSE) into: i) the efficient (true) price component and ii) a pricing error due to microstructure. The pricing error was on average 0.33% of the stock price.
8. Transparency
1.
“Market transparency” can be divided into:
Ex-ante transparency: the available market information on trading interest (quotes and orders, but not trades) Ex-post transparency: the available market information on executed trades (information on how money actually changed hands)
2.
1.
Ex-ante transparency:
Dealer markets: information available on i) quotes, ii) dealer identity, and possibly also iii) quantity at the bid and ask (if quotes not firm, the information is of course less reliable). LOB markets: the contents of the whole order-book visible for broker’s: i) all bid and ask orders, including ii) price, iii) quantity, iv) broker, v) time submitted, vi) type of order
2.
In a LOB, investors get the best information of the trading interest on both bid and ask side
8. Transparency
1.
Ex-post transparency:
Dealer markets: on OTC-markets (currency, interest rate etc.) no ex-post transparency, on exchanges information on executed trades with prices and volumes usually available LOB markets: information on executed trades with prices, volumes, buyer/seller identification, trade type etc.
2.
1.
Is transparency good or bad?:
Usually considered good for the market as a whole, since both ex-ante and ex-post transparency convey information to the market, increases fairness, keeps uninformed traders on the market and hence enhances liquidity and price discovery Transparency can be bad if informed traders think that they do not get rewarded for their superior information (which costs to acquire), and keep out of the market. This will reduce liquidity and harm price discovery which bad for all investors. On some markets, delayed trade information can be allowed to allow informed traders to benefit more from their information.
2.
9. OMX Helsinki
OMX Nordic Exchange:
1.
2.
3.
4.
5.
OMX Nordic Exchange consists of four local exchanges: Helsinki, Stockholm, Copenhagen and Iceland Separate legal entities in different jurisdictions: each local exchange has its own regulation The trading system and interface are the same (SAXESS), trading rules and practice harmonised but not exactly equal (e.g. trading hours not the same) Cross-membership for brokers easier once membership on one market is granted OMX also operates exchanges in Tallinn, Riga, and Vilnius
9. OMX Helsinki
Main features of OMX Helsinki:
1.
2.
3.
4.
5.
6. 7.
A limit order book market (LOB) Negotiated trades allowed (“Off-exchange registration”), must be reported to the exchange within 5 minutes. For very large blocks, reporting can be delayed to 60 minutes or even until the end of the day (trades above 5% of market cap.) Market making (Liquidity providing, “LP”) contracts allowed between broker and issuer (must be approved by exchange) “Automatic order routing” allowed for customers (must be accepted by the exchange) An order must contain information on if the order is made: i) for a customer, ii) own account, iii) issuer, iv) market making, v) stabilisation No round lots (one round lot is 1 share) Minimum tick size is 0.01 EUR for all shares, other OMX exchanges has stepwise ticks depending on share price
9. OMX Helsinki
Trading sessions:
1.
2.
3.
Pre-trading session (9.45 – 10.00?): a “morning call”, where the opening price is established as the trade-maximising call price. Trades not executed in the call are moved to the LOB for later continuous trading (if allowed by order type). Trading session: Continuous trading 10.00 – 18.20 Post-trading session (18.20 – 18.30): a “closing call” where the closing price is established.
9. OMX Helsinki
In July 2007, there were appr. 60 brokers The 10 largest by (EUR) market share:
Turnover Rank Member 1 2 3 4 5 6 7 8 9 10 Sign 56 41 39 33 33 30 27 27 26 25 783 243 644 330 137 234 440 101 977 226 105 680 400 568 837 221 861 336 847 625 EUR 650 569 841 957 810 463 057 350 268 103 % 8.23 5.98 5.75 4.83 4.81 4.38 3.98 3.93 3.91 3.66 Morgan Stanley & C o. International Ltd. MSI Deutsche Bank AG DBL Goldman Sachs International GSI Merrill Lynch International MLI C redit Suisse Securities (Europe) Ltd C SB C itigroup Global Markets Limited SAB Nordea Bank Finland Plc NRD UBS Limited UBS Lehman Brothers International LBI C arnegie Investment Bank AB C AR Source: www.omxgroup.com
Trades No 1 986 360 1 338 756 967 397 874 036 1 396 910 681 971 1 461 365 929 767 1 594 751 512 900 % 8.59 5.79 4.18 3.78 6.04 2.95 6.32 4.02 6.89 2.22
Appendix A: Some links for practical information
Global statistics:
www.world-exchanges.com
Equity markets:
www.omxgroup.com www.nyse.com www.nasdaq.com www.deutsche-boerse.com www.euronext.com
Derivative markets:
www.euronext.com www.eurexchange.com www.cboe.com