Financial Market Microstructure - HEC Paris...The Market Microstructure Approach It recognizes the...
Transcript of Financial Market Microstructure - HEC Paris...The Market Microstructure Approach It recognizes the...
Financial Market Microstructure
Stefano Lovo
HEC, Paris
Illiquidity and Price Impact
Standard approach in Financial Economics (e.g. CAPM,Asset pricing, Derivatives pricing etc . . . ): investors can tradeany quantities at equilibrium prices.
Law of one price: Two identical assets must trade for thesame price. The same asset must trade for the same pricein all locations.
Informational efficiency of the stock market: the price of anasset reflects all information about the asset’sfundamentals.
Stefano Lovo, HEC Paris Key concepts 2 / 1
Market Informational Efficiency
DefinitionWeak form efficiency: Trading prices incorporate all pastpublic information.Semi-Strong form efficiency: Trading prices incorporateall present and past public information.Strong form efficiency: Trading prices incorporate allpublic and private information available in the economy.
Stefano Lovo, HEC Paris Key concepts 3 / 1
Actual markets: low of one price?
US treasury bills: bonds with maturity of less than 2 years.US treasury notes: bonds with maturity of more than 2 years.
At any time there are coexist bill and notes with only oneresidual payment at the same future date.
Amihud and Mendelson (1991): On average notes trade at adiscount
notes’ average annual yield ' bills’ average annual yield +0.43%
Stefano Lovo, HEC Paris Key concepts 4 / 1
Actual markets: low of one price?
US treasury bills: bonds with maturity of less than 2 years.US treasury notes: bonds with maturity of more than 2 years.
At any time there are coexist bill and notes with only oneresidual payment at the same future date.
Amihud and Mendelson (1991): On average notes trade at adiscount
notes’ average annual yield ' bills’ average annual yield +0.43%
Stefano Lovo, HEC Paris Key concepts 4 / 1
Actual markets: price reflects fundamentals?
Stefano Lovo, HEC Paris Key concepts 5 / 1
Market liquidity
DefinitionAn asset’s liquidity is a measure of
The speed at which an asset can be bought and sold. (thefaster, the more liquid)
The price impact of the traded quantity. (the smallerimpact, the more liquid)
The cost of a "round-trip". (the cheapest, the more liquid)
Stefano Lovo, HEC Paris Key concepts 6 / 1
Why do we care?
Illiquidity means lower returns on portfolios: (i)portfolio managers care about market liquidity and (ii) thebrokerage industry devises trading strategies to minimizecosts due to illiquidity.
Illiquidity affects asset prices/cost of capital: (i)illiquidity = "tax" on asset payoffs + (ii) source of risk.
Price discovery affects the allocation of capital in theeconomy.
Stefano Lovo, HEC Paris Key concepts 7 / 1
An example on liquidity and cost of capital
Mid quote =(18.25 + 18.375)/2 = 18.3125Transaction price ' 18.3125± 0.0625transaction cost ' 0.34% of 18.3125
Example
Stock A is an immediate perpetuity paying $1 per year. The requiredreturn of capital is r = 5% per year. What is the current price for stockA? Answer $21
Suppose that at resale transaction costs are s = 0.34% of the price.What is the current price for stock A? Answer $19.66 ' $21(1−6.4%)
Stefano Lovo, HEC Paris Key concepts 8 / 1
Main Questions
1 Liquidity: What are the determinants of market liquidity?Liquidity risk? Measures (bid-ask spreads and depth)?
2 Price Discovery: How and to what extent do pricesimpound new information? At which speed?
3 Volatility: What are the determinants of price changes atthe high frequency? How do volatility and liquidity interact?
Answers have implications for (i) asset pricing (illiquiditypremium), (ii) the design of trading systems, (iii) regulation offinancial markets (best execution, insider trading, highfrequency trading ...), (iv) portfolio management (v) marketreactions to news, interventions etc...
Stefano Lovo, HEC Paris Key concepts 9 / 1
The Market Microstructure Approach
Some definitions (the term is originally due to Garman (1976))
O’Hara (1995) : "The study of the process and outcomesof exchanging assets under explicit trading rules..."
Madhavan (2000) : "The process by which investors?latent demands are ultimately translated into transactions".
Biais, Glosten and Spatt (2005) : "The investigation ofthe economic forces affecting trades, quotes and prices"
Stefano Lovo, HEC Paris Key concepts 10 / 1
Roadmap
Key conceptsAuctionsQuote driven markets: static models
Inventory modelsInformation models
Quote driven markets: dynamic modelsMarket efficiency and herd behaviorRobust price formation
Limit Order MarketsOTC markets
Stefano Lovo, HEC Paris Key concepts 11 / 1
The Market Microstructure Approach
It recognizes the role of:Heterogeneity among Market Participants. Participantsto security markets have various objectives (e.g. dealersare different from final investors; hedgers different fromspeculators etc...).
Institutional framework. Market design and marketregulation matter.
Private Information. Informational asymmetries amongmarket participants are prevalent in securities markets.
Stefano Lovo, HEC Paris Key concepts 12 / 1
Market Participants
Costumers
Dealers
Intermediaries
Stefano Lovo, HEC Paris Key concepts 13 / 1
Market Participants
Costumers: Agents that are willing to trade the security:Institutional investors (pension funds, mutual funds,foundations): Hold and manage the majority of assets;account for the bulk of trading volume; trade largequantities.Individual Investors (retail traders, household, banks):Account for the bulk of trades; trade smaller quantities.
Dealers
Intermediaries
Stefano Lovo, HEC Paris Key concepts 14 / 1
Market Participants
Costumers
Dealers: Large professional traders who do trade for theirown account and provide liquidity to the market.
Intermediaries
Stefano Lovo, HEC Paris Key concepts 15 / 1
Market Participants
Costumers
Dealers
Intermediaries:Brokers: Match costumer orders but do not trade for theirown account.Specialists (NYSE): Are responsible for providing liquidityand smoothing trade on given securities.Market Makers: Agents that stand ready to buy and sell thesecurity at their bid and ask prices respectively. Liquiditysuppliers.
Stefano Lovo, HEC Paris Key concepts 16 / 1
Institutional Framework: Market types
Call Auction Markets: Occur at specific time (ex. at theopening and or at the fixing); investors place orders thatare executed at a single clearing price that maximizes thevolume of trade.
Continuous Auction Markets (or limit order markets):Investors trade against resting orders placed earlier byother investors (Euronext, Toronto SE, ECNs).
Dealer Markets (or quote driven markets) :Market-makers post bid and ask quotes at which investorscan trade. (Bond Markets MTS, FX markets).
Alternative Trading Systems (ATS)Electronic Communication Networks (ECN): Continuousorder driven anonymous markets (Island, Instinet,Archipelago, Redibook)Crossing Networks (CN): Cross multiple orders at a singleprice determined in a base market. (POSIT, Xtra XXL)
Stefano Lovo, HEC Paris Key concepts 17 / 1
Most common orders
Buy limit order: Order to buy up to a quantity q for a pricenot larger than p.Sell limit order: Order to sell up to a quantity q for a pricenot smaller than p.
Buy market order: Order to buy a quantity q at the bestcurrent market conditions.Sell market orders: Order to buy a quantity q at the bestcurrent market conditions.
Stefano Lovo, HEC Paris Key concepts 18 / 1
A game theoretical approach
The market for one financial asset can be represented asfollows:
Market Participants: set of players NInstitutional framework:
Set of actions Xi available to market participant iAsset allocation rule Q : ×i∈NXi → RN
Cash allocation rule P : ×i∈NXi → RN∑i
Qi =∑
i
Pi = 0
Asset fundamental value vParticipant i ’s monetary payoff from transaction:
vQi + Pi
Participant i ’s utility after transaction, given initial wealthWi :
Ui(Wi + vQi + Pi)
Stefano Lovo, HEC Paris Key concepts 19 / 1
Modelling uncertainty
Ω: set of all possible states of Nature (finite).A: Set of all subsets of Ω.π : A → [0,1]: probability measure of A.
v : Ω→ R: Value of the asset (v ).Wi : Ω→ R: Value of agent i initial portfolio (Wi ).
Stefano Lovo, HEC Paris Key concepts 20 / 1
Partitions
DefinitionA partition P of Ω is a collection of nonempty, pairwisedisjoint subsets of Ω whose union is Ω.P(ω) denotes the element of P that contains the state ω.
Let P and P ′ be two partitions of Ω.
DefinitionPartition P is said to refine partition P ′ if every element θ of P iscontained in some element θ′ of P ′.
DefinitionThe meet of the partitions Pi and Pj , that we will denoteMij , isthe finest partition that is refined by both Pi and Pj .
Stefano Lovo, HEC Paris Key concepts 21 / 1
Modelling incomplete information using partitions
All agents start with common prior π over Ω.Agent i receive private information that we described as apartition Pi over Ω:if the true state is ω, then agent i is informed that the truestate belongs to the set Pi(ω).If the true state is ω, the type of agent i is θi = Pi(ω).
Stefano Lovo, HEC Paris Key concepts 22 / 1
An example
Ω = ω1, ω2, ω3, ω4, π(ω) = 0.25v(ω1) = 2; v(ω2) = 3, v(ω3) = 3, v(ω4) = 4P1 = ω1, ω2, , ω3 ω4P2 = ω1, ω3, , ω2 ω4P3 = ω1, ω2, , ω3, ω4P3 = ω1, ω2, ω3, ω4
Agent 1 and 2 receive different partial information; Agent 3is perfectly informed; Agent 4 is perfectly uninformed.
If for example ω = ω3, thenfor agent 1 E [v |P1(ω3)] = 3.5for agent 2 E [v |P2(ω3)] = 2.5for agent 3 E [v |P3(ω3)] = 3for agent 4 E [v |P4(ω3)] = 3
Stefano Lovo, HEC Paris Key concepts 23 / 1
Bayesian Equilibrium
Let Pi = θ1i , . . . θ
ki be the set of possible types for agent i .
A strategy for player i is a mapping
σi : Pi → ∆Xi
A Bayesian Nash equilibrium is a strategy profileσi , . . . , σN such that
σi(θi) ∈ arg maxxi∈Xi
E [Ui(Wi + vQi(xi , σ−i) + Pi(xi , σ−i))|θi ,h]
where h represents what is common knowledge at the timeof trading.
Stefano Lovo, HEC Paris Key concepts 24 / 1
Common knowledge
Something is common knowledge if we both know that it’strue;
1 and I know that you know it’s true;2 and you know that I know it’s true;3 and I know that you know that I know that you know that I
know that you know it’s true;. . .
4 and so on, for any string of beliefs we put together.
More formally,
LemmaIf the true state is ω, then what is common knowledge for playeri and j isMij(ω).
Stefano Lovo, HEC Paris Key concepts 25 / 1
Agreen to Disagree
Theorem( Aumann 1976) If two agents have the same priors, and afterreceiving private information their posterior on an event a ∈ Aare common knowledge, then these posteriors are equal.
Corollary
Take anyMij(ω). If the agents i and j have the same priorsthen for all ω′ ∈Mij(ω), we have
πi(ω′|Mij(ω)) = πj(ω
′|Mij(ω)),
i.e. posterior beliefs conditional on a common knowledge eventdo not differ.
Stefano Lovo, HEC Paris Key concepts 26 / 1
Trade Motivation
There are two possible reasons for trading:
Speculate on private information about v .Hedging, when Ui is concave.
Stefano Lovo, HEC Paris Key concepts 27 / 1
No trade theorem (simplified version)
TheoremConsider two agents with common prior, with the same weaklyconcave utility function U and with the same initial wealth W .Suppose each trader receives some private information. Then
In equilibrium there cannot occur any trade that allows atleast one trader to be strictly better off.If the utility function is strictly concave, then in equilibriumthere is no trade.
Stefano Lovo, HEC Paris Key concepts 28 / 1
No trade theorem (Migrom and Stokey 1982)
TheoremIf traders start from common priors and it is commonknowledge that all traders are rational and the current allocationis ex-ante Pareto efficient, then new asymmetric information willnot lead to trade, provided that traders are strictly risk averse.
CorollaryIf traders start from common priors and have no reason to tradea priori, then they will not trade based on the arrival of newprivate information.
Stefano Lovo, HEC Paris Key concepts 29 / 1
Proof (simplified version)
Suppose that there is a trade leading agent 1 to trade Q1asset for P1 dollars.The resulting change in i ’s wealth from the trade in state ωis
T (ω) := v(ω)Q1 + P1
Agent 1’s post trade utility is
U(W + T )
The resulting change in 2’s wealth from the trade in state ωis −T (ω).Agent 1’s post trade utility is
U(W − T )
Stefano Lovo, HEC Paris Key concepts 30 / 1
Proof (simplified version)
Suppose agent 1 is of type θ1 and he is willing to agree on thetrade.
Conditional on agent 2 agreeing, agent 1’s posterior belief onagent 2’s type being θ2 is
π(θ2|θ1, trade)
Thus, agent 1 evaluates the trade at∑θ2∈P2
π(θ2|θ1, trade)E[U(W + T )|θ1, θ2
]
If agent 1 strictly prefers trading, this is strictly larger than∑θ2∈P2
π(θ2|θ1, trade)E[U(W )|θ1, θ2
]
Stefano Lovo, HEC Paris Key concepts 31 / 1
Proof (simplified version)
Multiplying by π(θi , trade) :∑θ2∈P2
π(θ1, trade)π(θ2|θ1, trade)E[U(W + T )|θ1, θ2
]=
∑θ2∈P2
π(θ1, θ2, trade)E[U(W + T )|θ1, θ2
]>
∑θ2∈P2
π(θ1, θ2, trade)E[U(W )|θ1, θ2
]Suming across θ1:∑
θ1∈P1
∑θ2∈P2
π(θ1, θ2, trade)E[U(W + T )|θ1, θ2
]=
E[U(W + T )|trade
]> E
[U(W )|trade
]Because agent 1 strictly prefers trading for some θ1 one has
Stefano Lovo, HEC Paris Key concepts 32 / 1
Proof (simplified version)
Let Ω+ (resp. Ω−)be the set of states ω where agent 1 gains (resp. does not gain) from trading:
Ω+ := ω ∈ Ω|T (ω) > 0
Ω− := ω ∈ Ω|T (ω) ≤ 0
Then E[U(W + T )|trade
]> E
[U(W )|trade
]implies
∑ω∈Ω+
π(ω|trade)(U(W (ω) + T (ω))− U(ω)) >∑
ω∈Ω−π(ω|trade)(U(ω)− U(W (ω) + T (ω)))
Because U is weakly concave:
∑ω∈Ω+
π(ω|trade)(U(ω)− U(W (ω)− T (ω))) ≥∑
ω∈Ω+
π(ω|trade)(U(W (ω) + T (ω))− U(ω))
∑ω∈Ω−
π(ω|trade)(U(ω)− U(W (ω) + T (ω))) ≥∑
ω∈Ω−π(ω|trade)(U(W (ω)− T (ω))− U(ω))
These three inequalities together imply
E[U(W − T )|trade
]< E
[U(W )|trade
]that is player 2 will not agree to trade.
Stefano Lovo, HEC Paris Key concepts 33 / 1
Summary
1 Financial markets display frictions and illiquidity.2 Market microstructure: The investigation of the economic
forces affecting trades, quotes and prices.3 Market Participants.4 Market Mechanism.5 Asymmetric information.6 No trade theorem: trade cannot purely for speculation on
private information.
Stefano Lovo, HEC Paris Key concepts 34 / 1
References
Yakov Amihud and Haim Mendelson (1991); “Liquidity,maturity, and teh yields on U.S. treasury securities”,Journal of Finance.Robert Aumann (1976), “Agreeing to Disagree” Annals ofStatistics.John J. Geanakoplos and Herakles Polemarchakis (1982),“We Can’t Disagree Forever,” Journal of Economic Theory.Paul Milgrom and Nancy Stokey (1982), “Information,Trade and Common Knowledge,” Journal of EconomicTheory.
Stefano Lovo, HEC Paris Key concepts 35 / 1