Liquidity and Information in Order Driven Markets

Ioanid Roşu∗

February 25, 2016

Abstract

How does informed trading affect liquidity in order driven markets, where
traders can choose between market orders (demanding liquidity) and limit orders
(providing liquidity)? In a dynamic model of order driven markets we find that
informed trading overall helps liquidity: a higher share of informed traders (i)
improves liquidity as proxied by the bid-ask spread and market resiliency, and (ii)
has no effect on the price impact of orders. The model generates other testable
implications, and suggests new measures of informed trading.

JEL: C73, D82, G14

Keywords: Limit order book, volatility, trading volume, slippage, informed

trading, stochastic game.

∗
HEC Paris, [email protected], +33 6 72 09 76 02. The author thanks Peter DeMarzo, Doug Diamond,
Thierry Foucault, Johan Hombert, Peter Kondor, Juhani Linnainmaa, Christine Parlour, Uday Rajan,
Stefano Lovo, Tālis Putniņš, Pietro Veronesi for helpful comments and suggestions. He is grateful to
finance seminar participants at Chicago Booth, Stanford University, University of California at Berkeley,
University of Illinois Urbana-Champaign, University of Toronto (Dept. of Economics), Bank of Canada,
HEC Lausanne, HEC Paris, University of Toulouse, Ecole Polytechnique, Tilburg University, Erasmus
University, Insead, Cass Business School; and to conference participants at the 2010 Western Finance
Association meetings, 2010 European Finance Association meetings, NBER microstructure meeting, 4th
Central Bank Microstructure Workshop, and 1st Market Microstructure Many Viewpoints Conference
in Paris.

1
1 Introduction
Market liquidity is a central concept in finance, in particular in relation with asset
pricing.1 According to Bagehot (1971), illiquidity is caused by asymmetric information,
via the actions of liquidity providers. The liquidity provider, or market maker, which
Bagehot identifies as the “exchange specialist in the case of listed securities and the
over-the-counter dealer in the case of unlisted securities,” sets prices and spreads so
that on average he makes losses from traders who possess superior information, but
compensates with gains from uninformed traders, who are motivated by liquidity needs
or simply trade on noise. Thus, the stronger the asymmetric information between the
informed traders and the market maker, the larger the bid-ask spread needs to be so
that the market maker at least breaks even. A large theoretical literature has since
made Bagehot’s intuition rigorous.2
Following Bagehot (1971), most of the theoretical literature assumes that liquidity
providers do not to possess any superior fundamental information.3 More recent ev-
idence, however, has called into question this assumption. One reason is that most
financial exchanges around the world have become “order driven,” meaning that any in-
vestor (informed or not) can provide liquidity by posting orders in a limit order book.4
Moreover, empirical evidence shows that there is an important premium for liquidity
provision in order driven markets, and that informed traders do indeed use limit orders
extensively.5 Despite the evidence, the literature has been largely silent on the order
choice problem of informed traders, and, importantly, on how this choice affects market
liquidity. The goal of the present paper is to fill this gap.
We thus consider the following set of questions: What is the optimal order choice of
1
See Amihud and Mendelson (1986), Brennan and Subrahmanyam (1996), Easley, Hvidkjaer, and
O’Hara (2002), Pástor and Stambaugh (2003), Acharya and Pedersen (2005).
2
See Kyle (1985), Glosten and Milgrom (1985), or O’Hara (1995) and the references within.
3
Notable exceptions are Chakravarty and Holden (1995), Kaniel and Liu (2002), and Goettler,
Parlour, and Rajan (2009).
4
Nowadays, most equity and derivative exchanges are either pure order driven markets (Euronext,
Helsinki, Hong Kong, Tokyo, Toronto); or hybrid markets, in which designated market makers must
compete with a limit order book (NYSE, Nasdaq, London). See Jain (2005).
5
See Biais, Hillion, and Spatt (1995), Harris and Hasbrouck (1996), Griffiths, Smith, Turnbull, and
White (2000), Sandås (2001), Hollifield, Miller, and Sandås (2004), Anand, Chakravarty, and Martell
(2005), Rourke (2009), Menkhoff, Osler, and Schmeling (2010), Latza and Payne (2011), Hautsch and
Huang (2012). Similar findings are reported by Bloomfield, O’Hara, and Saar (2005) in the context of
experimental asset markets.

2
an informed trader in an order driven market? How does a larger fraction of informed
traders affect liquidity? What is the information content of limit orders and market
orders? How does the market recover after a liquidity or information shock? Can the
time series of market and limit orders be used to infer the fraction of informed trading?
To address these questions, we consider a dynamic model of an order driven market.
Risk-neutral investors arrive randomly to the market and trade in one risky asset. The
asset’s fundamental value is time varying, and information about it is costly to acquire
and process.6 Informed investors learn the current value of the asset, and decide whether
to buy or sell one unit of the asset, and whether to trade with a market order or a limit
order. Limit orders can subsequently be modified or cancelled without any cost.
Our main result is that a larger fraction of informed traders overall improves liquidity.
This result is driven by two key features of the model: First, there is competition among
informed traders, as each informed trader must take into account the future arrivals of
other informed traders. Second, private information is long-lived, in the sense that
information about the fundamental value is revealed to the public only via the order
flow.7 Because of these features, a larger share of informed traders produces a dynamic
efficiency that can eventually overcome the static increase in adverse selection. To
understand in more detail the intuition behind our main result, we briefly describe
several key equilibrium results.
The first key result describes the optimal order choice of the informed trader. This is
essentially a threshold strategy: the informed trader (referred to in the paper as “she”)
submits either a market order or a limit order, depending on the magnitude of her
privately observed mispricing, which is the difference between the fundamental value
(privately observed) and the efficient price (the public expectation of the fundamental
value). An extreme mispricing causes the informed trader to submit a market order,
while a moderate mispricing causes a limit order. This result formalizes an intuition
present for instance in Harris (1998), Bloomfield, O’Hara, and Saar (2005), Hollifield,
6
Because we are interested in long run liquidity effects, we assume that the asset value is not
constant, but follows a random walk. Thus, prices do not eventually reveal all the private information.
In Goettler, Parlour, and Rajan (2009), the fundamental value is also time varying, but follows a
Poisson process.
7
Goettler, Parlour, and Rajan (2009) obtain different results because in their model the private
information is short-lived (the fundamental value is revealed publicly after several periods).

3
Miller, Sandås, and Slive (2006), Large (2009).
The second key result describes the information content of the order flow. Because
in equilibrium informed traders can submit both limit orders and market orders, all
types of order have price impact (defined as the change in efficient price caused by the
order). Nevertheless, because market orders are associated to more extreme mispricing,
the price impact of a buy market order is larger in magnitude (about 4 times larger
in our model) than the price impact of a buy limit order. In line with this prediction,
Hautsch and Huang (2012, p.515) find empirically that market orders have a permanent
price impact of about four times larger than limit orders of comparable size.
Our third key result describes the equilibrium bid-ask spread, and identifies a new
component of this spread: the slippage component. We define slippage as the tendency
of an informed trader’s estimated mispricing to decay over time.8 Slippage is due to
the future arrival of other informed traders who correct the mispricing by submitting
their orders. Thus, slippage induces an endogenous waiting cost for the informed trader,
called the slippage cost. In addition, the informed trader suffers from an adverse selection
cost, since at the time of order execution she is potentially less informed than the future
informed traders.9 We define the decay cost as the sum of the slippage cost and the
adverse selection cost.
The decay cost generates a tradeoff between limit orders and market orders: by
trading with a limit order, an informed trader gains half the bid-ask spread, but loses
from the decay cost. By trading with a market order instead, the informed trader
loses half the bid-ask spread, but pays no decay cost. At the threshold mispricing,
the informed trader is indifferent between a market order and a limit order. Hence,
the decay cost corresponding to this threshold value is equal to the equilibrium bid-
ask spread. From the definition of the decay cost, the bid-ask spread is therefore the
8
Finance practitioners sometimes refer to “slippage” with a different meaning than in our paper.
For instance, Investopedia refers to the slippage of a large (potentially uninformed) market order, for
which each additional unit executes at a worse price—a phenomenon also known as walking the book.
In contrast, in our model slippage applies only to limit orders submitted by informed traders, and it
occurs even when limit orders are for just one unit.
9
This is because the informed trader acquires information only when she enters the market. If
instead she continuously observes the fundamental value, the adverse selection component is zero, but
the slippage cost is still positive, as competition with future informed traders gradually erodes her
initial information advantage.

4
sum of a slippage component and an adverse selection component. To our knowledge,
the slippage component is new to the literature. Huang and Stoll (1997) decompose the
bid-ask spread into order processing costs, adverse selection costs, and inventory holding
costs. In our model, we abstract away from inventory issues and order processing costs,
but recover the adverse selection component. In addition, however, we show that by
allowing informed traders to provide liquidity, the phenomenon of slippage generates a
new component of the bid-ask spread.
Our main result describes how liquidity is affected by the fraction, or share of in-
formed traders, henceforth called the informed share. Surprisingly, a larger informed
share overall has a positive effect on liquidity.10 More precisely, a larger informed share
has (i) a negative effect on bid-ask spreads; (ii) no effect on the price impact; and (iii) a
strongly positive effect on market resiliency (define in Kyle 1985 as the speed with which
prices recover from a random, uninformative shock). Moreover, a larger informed share
has a positive effect on market efficiency by reducing the efficient volatility. The latter
is defined as the publicly inferred volatility of the fundamental value, hence its inverse
is a measure of dynamic efficiency: when the efficient volatility is small, the public has
precise information about the fundamental value.
To get intuition for the main result, note that a larger informed share implies that
the informed traders exert more pressure on prices to revert to the fundamental value.
This explains the strong positive effect of the informed share on market resiliency. Also,
it explains the negative effect of the informed share on efficient volatility: when there
are more informed traders, the public eventually learns better about the fundamental
value, and the efficient volatility decreases. But the bid-ask spread is equal to the decay
cost corresponding to the threshold mispricing. When the efficient volatility is smaller,
the decay cost is also smaller because the average mispricing tends to be smaller. Hence,
a larger informed share generates a smaller bid-ask spread.
To understand the neutral effect of the informed share on market depth, suppose
10
Several empirical papers document this positive relationship, although the interpretation offered
is different from ours. Collin-Dufresne and Fos (2013) find that the bid-ask spread and realized price
impact decrease in the presence of corporate insider trading. In our model, price impact is not affected
by informed trading, but this might be due to the fact that we measure the instantaneous price impact,
while they estimate the realized price impact, which might be affected by market resiliency when
investors are strategic. Brennan and Subrahmanyam (1995) find that more analyst coverage for a
security improves its realized price impact.

5
the informed share is small, and a buy market order arrives. There are two opposite
effects at play. First, when the informed share is small, it is unlikely that the market
order comes from an informed trader. This effect decreases the price impact. But,
second, if the buy market order does come from an informed trader, she must have
observed a fundamental value far above the efficient price; otherwise, knowing there is
little competition from other informed traders, she would have submitted a buy limit
order. This effect increases the price impact. The two effects exactly offset each other.11
The results described thus far assume a special type of equilibrium we call homothetic,
by which we mean that the efficient volatility is constant over time (which in turn makes
the bid-ask spread and price impact also constant). In the homothetic equilibrium, the
natural increase in uncertainty due to changes in the fundamental value is exactly offset
by the new information contained in the order flow. Our final set of results arise from the
study of non-homothetic equilibria, which can appear for instance after an uncertainty
shock (an unobserved shock to the fundamental value) results in a temporary spike in
efficient volatility.
We find that liquidity is resilient in our model: after an uncertainty shock, the bid-
ask spread and price impact (as well as the efficient volatility) decrease over time to
their values in the homothetic equilibrium. The bid-ask spread and price impact are
both increasing in the size the uncertainty shock. The liquidity resilience is larger when
there are more informed traders, as the order flow becomes more informative. Liquidity
resiliency is different from market resiliency, as the latter is the tendency of prices to
revert to the fundamental value after an uninformative shock.
We introduce a new measure, the market-to-limit probability ratio, which is the de-
fined as the probability the next order is a market order, divided by the probability
that the next order is a limit order. This number is equal to one in the homothetic
equilibrium, but after an uncertainty shock the market-to-limit probability ratio drops
to levels significantly less than one, as the increase in the bid-ask spread temporarily
prompts the informed traders to submit more limit orders. The connections among the
market-to-limit probability ratio with the liquidity measures and the efficient volatility,
as well as the expected evolution of the equilibrium towards the homothetic one, produce
11
This is proved rigorously in Proposition 1, and explained in the subsequent discussion.

6
new testable implications of the model.
Overall, our theoretical model produces a rich set of implications regarding the con-
nection between the activity of informed traders and the level of liquidity. We find
that informed traders have on aggregate a positive effect, by making the market more
efficient and, at the same time, more liquid. A welfare analysis in Internet Appendix
Section 3 also shows that a larger number of informed traders (caused for instance by an
exogenous decrease in information costs) increases aggregate trader welfare. Our model
thus provides useful tools to analyze informed trading, and its connection with liquidity,
prices, and welfare.
Our paper is part of a growing theoretical literature on price formation in order
driven markets.12 Of central interest in this literature is how investors choose between
demanding liquidity via market orders and supplying liquidity via limit orders.13 Several
papers, such as Foucault, Kadan, and Kandel (2005), or Roşu (2009) study order choice
by assuming that investors have exogenous waiting costs. One advantage of our model
is that waiting costs arise endogenously in the case of an informed investor: these are
the decay costs described above.
Goettler, Parlour, and Rajan (2009) is the first paper that solves a dynamic model of
order driven markets with asymmetric information. The focus of their paper is however
different than ours. While we are interested in the effect of informed trading on liq-
uidity, Goettler, Parlour, and Rajan (2009) analyze the interplay between information
acquisition, order choice and volatility. They find that under picking off risks—which
are absent in our model—different volatility regimes affect traders’ order choice, and
make the market act as a volatility multiplier. Moreover, there are two important mod-
eling differences. First, in their model private information is short-lived, because the
fundamental value is publicly revealed after several periods. This assumption reduces
the effect of dynamic efficiency in their model, as informed traders cannot arrive more
quickly to make the market more efficient. By contrast, in our model dynamic efficiency
12
See Glosten (1994), Parlour (1998), Foucault (1999), Foucault, Kadan, and Kandel (2005), Goet-
tler, Parlour, and Rajan (2005, 2009), Back and Baruch (2007), Roşu (2009), Biais, Hombert, and Weill
(2013), Pagnotta (2010), and the survey by Parlour and Seppi (2008).
13
For models of order choice without private information about the fundamental value, see Cohen,
Maier, Schwartz, and Whitcomb (1981), Harris (1998), Foucault (1999), Parlour (1998), Goettler,
Parlour, and Rajan (2005), Roşu (2009).

7
has a strong effect by having private information being incorporated over the long run,
and as a result the informed traders have an overall positive effect on liquidity. Second,
in their model traders do not continuously monitor the market, which creates stale limit
orders and picking off risks.14 In our model, there are no stale orders since limit orders
can be modified instantly.
The paper is organized as follows. Section 2 describes the model. Section 3 solves for
the homothetic equilibrium, in which the efficient volatility (as well as the bid-ask spread
and price impact) is constant. Section 4 describes the properties of the homothetic
equilibrium, including the various dimensions of liquidity and information efficiency.
Section 5 explores non-homothetic equilibria of the model. Section 6 concludes. Proofs
of the main results are in the Appendix and the Internet Appendix. The companion
Internet Appendix contains additional results and robustness checks.

2 Model
We consider a dynamic model of trading in a single asset. Time is continuous and traders
arrive randomly to the market. After deciding whether to acquire private information
regarding the fundamental value of the asset, traders can submit an order to buy or sell
one unit of the asset. Traders also choose the price at which they are willing to transact.
If an order does not execute, it can be subsequently modified or cancelled. Information
can be difficult to process, in a sense that is made precise below. We now describe the
model in more detail.
Trading and Prices. The market mechanism is order driven, meaning that a
transaction takes place when a buy or sell order is executed against an order on the
opposite side. Each order is a limit order, as it specifies a quantity and a price beyond
which the trader is no longer willing to transact. The price can be any real number.
Limit orders are subject to price priority: buy orders submitted at higher prices and
sell orders submitted at lower prices have priority. Limit orders submitted at the same
price are subject to time priority: the earlier order is executed first. If several orders
14
Linnainmaa (2010) finds that limit orders are often stale in the presence of public news.

8
arrive at the same time, a random order is assigned to them.15
The limit order book is the collection of all outstanding limit orders (submitted but
not yet executed or cancelled). In the book, limit orders form two queues, based on
order priority: the ask queue on the sell side, and the bid queue on the buy side. The
lowest price on the ask side is the ask price, or simply the ask. The highest price on
the bid side is the bid price, or simply the bid. A marketable limit order is a buy limit
order with a price above the ask, or a sell limit order below the bid. A marketable limit
order is executed immediately and is henceforth called a market order.
Traders and Arrivals. Traders arrive to the market according to a Poisson process
with parameter λ. Immediately after arrival, a trader chooses whether to (a) submit a
market order, (b) submit a limit order, or (c) submit no order at all. Each order is for
one unit of the asset. After submission, a limit order can be either (i) modified, which
means the limit price is changed—in which case time priority is lost, or (ii) cancelled.
As soon as the order is executed or cancelled, or if no order is submitted, the trader
exits the model.
Traders are risk-neutral but their utility also includes a private valuation component
and a cost from waiting.16 Each trader has a type (u, r), which consists of a private
valuation u for the asset and a waiting coefficient r. The private valuation u can take
three possible values, {−ū, 0, ū}, where ū > 0. A trader is a natural buyer if u = ū, a
natural seller if u = −ū, or speculator if u = 0. At time t, the instantaneous utility of a
trader with private valuation u is

 v − pt + u, if trader buys at t,
 t


pt − vt − u, if trader sells at t, (1)



 0, if trader’s order does not execute at t,

where vt is the fundamental value at t, and pt is the transaction price at t. Traders

incur a waiting cost of the form r × τ , where τ is the expected waiting time, and r is a
constant coefficient. The waiting coefficient r can take two possible values, {0, r̄}, where
15
With Poisson arrivals, the probability of two or more traders arriving at the same time is zero.
16
The private valuation can arise from liquidity or hedging needs, or from bias regarding the asset
(optimism or pessimism). The waiting cost can arise from trading horizon/deadlines, or from uncer-
tainty regarding future order execution.

9
r̄ > 0. A trader is patient if r = 0, or impatient if r = r̄.
To simplify presentation, we assume that (i) impatient natural buyers always submit
a buy market order, (ii) impatient natural sellers always submit a sell market order, and
(iii) impatient speculators do not submit any order. In Internet Appendix Section 2,
we show that (i)-(iii) are equilibrium results if ū and r̄ are above certain thresholds.17
Since traders who submit no order exit the model immediately, we replace (iii) by the
assumption that all speculators are patient.
Natural buyers and sellers (traders with valuation ū or −ū) arrive randomly to
the market according to an independent Poisson process with parameter λu . They
are equally likely to have positive or negative private valuation, and equally likely to
be patient or impatient. Patient speculators arrive randomly to the market according
to an independent Poisson process with parameter λi . The total trading activity is
λ = λu + λi .
Information. At any time t, the asset has a fundamental value vt , also called
common value or full-information price. The asset value follows a diffusion process
dvt = σv dBt , where Bt is a standard Brownian motion, and the fundamental volatility
parameter σv is a positive constant. Because traders arrive to the market according to
a Poisson process, inter-arrival times are exponentially distributed with mean 1/λ. For
simplicity of notation, throughout the paper we work in event time rather than calendar
time: if a trader arrives at t, the next trader arrives at t+1.18 The discrete version of the
fundamental value process in event time is vt+1 = vt + σεt+1 , where εt+1 has a standard
√
normal distribution N (0, 1), and the inter-trade volatility parameter is σ = σv / λ.
By paying an information acquisition cost, a trader learns the fundamental value at
the time of arrival.19 To simplify presentation, we assume that only the patient specu-
lators acquire information; this is proved as an equilibrium result in Internet Appendix
Section 3. In what follows, we refer to the patient speculators as informed traders, and
17
In particular, we show that it is not profitable for a sufficiently impatient speculator to acquire
information. Ex post, after seeing the signal, such a speculator might observe an extreme mispricing
that could be exploited without waiting and would therefore justify the information cost, but ex ante
such signals are rare and therefore do not justify the cost.
18
This use of event time has been justified empirically for instance by Hasbrouck (1993). Equivalently,
we set the model in discrete time, in which case t + 1 is replaced by t + λ1 .
19
Alternatively, we can assume that the informed traders observe the asset value continuously, but
this does not simplify the solution, and we conjecture that the qualitative results are the same.

10
to the natural buyers and sellers as uninformed traders.
All traders observe the history of the game. As the market is order driven, the
history consists of the total order flow, i.e., submissions, executions, modifications, and
cancellations. The evolution of the limit order book and the transaction prices are part
of this public information. A trader’s type (private valuation and waiting coefficient)
is private information for each trader. The fundamental value at the time of arrival is
private information for each informed trader.
Equilibrium Concept. Our model represents a stochastic game, in which Nature
moves by drawing randomly new traders at each time t = 0, 1, 2, ... After traders arrive
and decide whether to become informed or not, they engage in a trading game and at
each time maximize their expected utility given their information set. Even though the
arrivals occur at discrete points in time, traders can later modify their orders at any time
in between. The game is therefore set in continuous time, and we use the framework of
Bergin and MacLeod (1993) in which traders can react instantly.
The equilibrium concept is Markov Perfect Equilibrium (MPE), as defined for in-
stance in Fudenberg and Tirole (1991). As a refinement of the Perfect Bayesian Equilib-
rium (PBE) concept, a MPE is defined by a game assessment, which is the collection of
a strategy profile and a belief system such that (i) at every stage of the game, strategies
are optimal given the beliefs, and the beliefs are obtained from equilibrium strategies
and observed actions using Bayes’ rule, and (ii) the game assessment is conditional on
a set of state variables which are payoff-relevant. The latter condition implies that in a
MPE there are no ad-hoc punishments to support the equilibrium.
Information Processing. Solving the model described above is very challenging if
traders can do full Bayesian updating. This is because each trader’s inference problem
involves an infinite number of state variables, which are the moments of the probability
density ψ that describes the trader’s belief about the fundamental value. As new orders
arrive, the belief ψ must be updated based the information contained in each order type.
But because informed traders use threshold strategies (see Theorem 1), the update of
the density ψ changes its shape in ways which are difficult to quantify precisely.
Our modeling approach is to introduce frictions in information processing such that

11
traders solve a simplified inference problem.20 These frictions are based on the principle
that it is more difficult to process (i) private rather than public information, (ii) condi-
tional rather than unconditional information, and (iii) higher rather than lower moments
of a distribution. But rather than explicitly introducing information processing costs,
we directly specify what information traders can process.
When updating the belief density ψ, an uninformed trader can compute without cost
(i) the first moment of the posterior belief conditional on order flow, and (ii) the un-
conditional second moment of the posterior belief. Uninformed traders cannot compute
higher moments, and their beliefs are always normally distributed. To avoid different
beliefs among uninformed traders, we assume that the initial belief of an uninformed
trader is such that after submitting a limit order in the direction of his private valuation,
his posterior belief coincides with the posterior belief of the other uninformed traders.21
Thus, the uninformed traders waiting in the order book have the same normally dis-
tributed belief ψ E , the efficient density. The efficient density is public knowledge. Its
mean is the efficient price v E , and its standard deviation is the efficient volatility σ E .
Private information is more difficult to process. An informed trader cannot update
the belief density ψ conditional on the order flow. But she can compute without cost the
first moment of the relevant payoff variable, which is the asset mispricing (the difference
between the asset value and the efficient price) at the time when her order is executed,
provided that she follows the strategy of an uninformed trader after she makes the
initial order choice. Equivalently, the informed trader makes the following decisions:
(i) at the time of arrival she chooses which order to submit, based on the expected asset
mispricing at the time of order execution, and (ii) if her choice is a limit order, she uses
an uninformed broker to later update the order until it is executed.
For tractability, we assume that an informed trader receives a small penalty ν if
after observing the fundamental value she chooses not to trade.22 This assumption is
20
Given the difficulty of the traders’ inference problem and the fact that information acquisition is
already costly in our model, it is plausible to assume that information processing is costly as well.
21
This assumption reconciles the divergence in beliefs that private knowledge about own type can
create. For instance, an uninformed trader who submits a limit order privately knows that his order
is uninformed, but the other uninformed traders do not know and may update their beliefs. See the
proof of Lemma A3 in the Appendix for a formal discussion.
22
This assumption in needed to avoid no-order regions for the informed trader, which can occur
when her perceived mispricing is close to zero.

12
equivalent to the informed trader receiving a private benefit ν if she submits an order to
the market, which intuitively can arise from “learning by trading.” Because ν indicates
a commitment to trade by the informed investor, we call it the commitment parameter.
In Section 5.2, we show that this assumption is necessary only if the number of informed
traders is above a threshold.
Robustness. The model described thus far can be solved essentially in closed form.
We therefore use it as a benchmark model to study the robustness of the equilibrium
results. In Internet Appendix Section 5 we study the effect of relaxing some of the
assumptions that are made for tractability. We then verify that the equilibrium is not
significantly affected by relaxing these assumptions.

2.1 Notation and Parameters

The exogenous parameters in the model are the fundamental volatility σv , the trading
activity λ, the uninformed trading activity λu , and the informed trading activity λi ,
√
subject to the equality λ = λu + λi . The inter-trade volatility is σ = σv / λ. The
investor preference parameters are the private valuation parameter ū, the impatience
parameter r̄, and the commitment parameter ν.
We now define four numeric parameters that are used extensively throughout the
paper. The first three are

1 φ(0) − φ(α)
α = Φ−1 3


4
≈ 0.6745, β = ≈ 0.7867, γ = ≈ 0.2554,
4φ(α) φ(α)
(2)
where φ( · ) is the standard normal density, and Φ( · ) is its cumulative density.
The fourth numeric parameter is the function g = g(ρ, w) from Definition A1 in
the Appendix. As explained in the discussion that follows equation (3), the function
g has an interpretation within our model. The definition itself, however, is completely
independent of this interpretation, and therefore g can be thought as a parameter.
Even though we have not been able to find a closed form expression for the information
function g, it can be estimated with good precision by using a numerical Monte Carlo
procedure described in detail in Internet Appendix Section 4.

13
 We introduce several more useful parameters:

λi
ρ = = informed share,
λi + λu
q σ
∆ = 2
1+γ 2
√v = impact parameter,
λ (3)
V = βρ−1 ∆, = volatility parameter,


S = α − g(ρ, α) V = spread parameter.

We now briefly explain how the information function g is interpreted in our model.
Recall that at any given date t, the uninformed traders regard the asset value vt as
distributed by the normal density ψtE (the efficient density), with mean vtE (the efficient
price) and volatility σtE (the efficient volatility). In the homothetic equilibrium of Sec-
tion 3, the efficient volatility is constant and equal to the parameter V from (3). It is
therefore convenient to normalize variables by V . For instance, we define the signal at
t to be the normalized mispricing

vt − vtE
wt = . (4)
V

Then for the uninformed traders the distribution of the signal in the homothetic equi-
librium remains the same for all t: wt ∼ N (0, 1), the standard normal distribution.
Consider an informed trader who, for simplicity, arrives at date t = 0 and observes
an asset value v0 when the efficient price is v0E . Then, g is a function of the informed
λi v0 −v0E
share ρ = λi +λu
and the initial signal w0 = V
. Suppose the informed trader submits
a buy limit order (BLO), after which she follows the strategy of an uninformed trader
(which is described in Corollary 4 below). Denote by T the random time when the BLO
is executed (by an SMO). Then, g(ρ, w0 ) is the expected signal wT conditional on the
BLO being executed at T by an SMO. Proposition 2 shows that this interpretation of
g is indeed correct.

14
3 Equilibrium
We summarize the trading game described in the previous section. At each integer time
t = 1, 2, . . . (corresponding to average arrival times 1/λ, 2/λ, . . .) the asset value changes
according to

σv
vt = vt−1 + σεt , with σ = √ and εt ∼ N (0, 1), (5)
λ

where σ is the inter-trade volatility, and εt has the standard normal distribution. At
each integer time t ≥ 0, Nature draws a new trader which is either (i) informed with
probability ρ, meaning that she observes vt , or (ii) uninformed with probability 1 −
ρ, in which case with equal probability the trader is either a patient natural buyer
(with private valuation ū), patient natural seller (with private valuation −ū), impatient
natural buyer (who always submits a BMO), or impatient natural buyer (who always
submits a SMO).23
A trader that arrives at t either chooses an order of type {BMO, BLO, SLO, SMO}
for one unit of the asset, or submits no order (NO). In the latter case, the trader
exits the model forever, and Nature immediately draws another trader from the pool.
After the initial order submission, a trader who has submitted a limit order uses an
uninformed broker to handle the subsequent limit order modifications or cancellations.
At non-integer times the game is played with the existing traders in the order book.
We are interested in Markov Perfect Equilibria (MPE) of the game. As with any
MPE, the traders’ strategies depend on a set of state variables. In our context, the
public state variables are (i) the efficient density, given by its first two moments, the
efficient price and the efficient volatility, and (ii) the limit order book, given by the ask
and bid prices, and the ask and bid queues. The private state variable is the asset value,
observed by each informed trader when arriving to the market.24
The MPE described in this section has the additional property that it is a homothetic
equilibrium, by which we mean that the efficient volatility remains constant over time.
In Section 5, we study non-homothetic MPEs corresponding to different initial values
23
The decisions of the impatient traders are endogenized in the Internet Appendix Section 2.
24
More details are given in the proof of Theorem 1 in the Appendix.

15
for the efficient volatility, and we show that over time all these equilibria approach the
homothetic MPE of this section.
Because we want the game to begin already in a homothetic equilibrium, we assume
that an instant before t = 0 the efficient price is zero and the efficient volatility is the
parameter V from (3), so that the initial efficient density is N (0, V 2 ). The ask price is
S/2, the bid price is −S/2, where S is the parameter from (3), while the initial order
book has countably many limit orders on each side (see the middle plot in Figure 2).

3.1 Intuition

Before we state formally our results, we describe intuitively the equilibrium of the model,
as well as the role played by several key assumptions.
To understand the behavior of an uninformed trader, consider a patient natural buyer
(with private valuation ū > 0 and zero waiting costs) who arrives at t = 0. Because
of his positive valuation, he prefers a buy order to a sell order, and because of his zero
waiting costs he also prefers a limit order to a market order.25 Then, after submitting
a BLO at t = 0, he waits for his order to be executed, and in the meantime he modifies
his bid to account for the information contained in the order flow.
For instance, suppose that a BMO is submitted at t = 1, when the efficient price
just before trading is v1E . With probability ρ, the BMO belongs to an informed trader
who, according to her equilibrium strategy, sees a fundamental value v1 above a certain
threshold (equal to v1E + αV , where α and V are the constant parameters defined in
Section 2.1). As a result of this new information, the uninformed traders raise the
efficient price by a positive amount which turns out to be ∆, the impact parameter
from (3). Moreover, because we are looking for a MPE in which the efficient price is a
state variable, it is natural to expect that in equilibrium all non-executed limit orders
in the book, including the initial BLO, are also shifted up by ∆. To complete the
description of the equilibrium, we also assume that the shift preserves the traders’ ranks
25
We conjecture that the main results in our paper remain robust to having small positive waiting
costs. We expect the solution of such a model, however, to be much more complicated. Indeed, as
seen in models of the limit order book with symmetric information but positive waiting costs, such as
Roşu (2009), the numbers of limit orders on each side of the book become additional state variables.

16
in the ask and bid queues.26
An additional requirement for the equilibrium is that the uninformed traders are
indifferent between being the first or higher in their ask or bid queue. For instance, ac-
cording to (1) the utility of an uninformed seller depends on expected difference between
the execution price pt and the fundamental value vt . If the SLO is executed at the ask
price pt = vtE + S/2, this difference turns out to be S/2 − ∆. Intuitively, a seller gains
S/2, which is half the equilibrium bid-ask spread, but loses ∆ from the adverse selection
caused by the BMO that executes his order. In this section, we look for a homothetic
MPE, in which both S and ∆ are constant, therefore the additional equilibrium condi-
tion is automatically satisfied. In a non-homothetic equilibrium, however, this condition
must be imposed (see Corollary 8 in Section 5).
A key simplifying assumption of our model is that information processing is costly,
and as a result uninformed traders always perceive the efficient density as normal. But
with perfect Bayesian updating, the informed trader’s threshold strategy would lead
to non-normal distributions. We therefore explore full Bayesian updating in Internet
Appendix Section 5.1, and we see that the departure from normality is small. We thus
conjecture that our main results remain true under perfect Bayesian updating.
Finally, to understand the behavior of an informed trader, it is enough to describe
her initial order submission.27 As explained in the introduction of the paper, this is
a threshold strategy. For instance, if she observes a fundamental value (vt ) above a
certain threshold depending on the efficient price (vtE + αV ), she optimally submits a
BMO. If instead she submits a BLO, she expects her information advantage to decay
over time; and the decay costs are larger when the mispricing she observes is larger (see
26
Rank preservation does not matter per se, as traders in the book have zero waiting costs. But if
we want the model to be more realistic, the rank in the queue is important, at least for the informed
traders. Indeed, one of our simplifying assumptions is that, after the initial order choice, the informed
traders hire an uninformed broker to handle their order execution. This raises a subtle issue: when
the informed traders compute their initial expected utility, they realize that their subsequent behavior
imitates that of the uninformed traders. But they could also realize that their average information
advantage decreases over time because of the future arrival of competing informed traders (this is
the phenomenon of “slippage” described in Section 4.3). Thus, the informed traders could instruct
their broker to jump ahead in the queue in order to ensure a faster order execution. To prevent this
behavior, we impose the out-of-equilibrium belief that jumping ahead in the queue can come only from
an informed trader.
27
Indeed, after their initial choice they become essentially uninformed, as their orders are subse-
quently handled by an uninformed broker.

17
Section 4.3). The benefit of a limit order relative to a market order, however, does not
depend on the observed mispricing, and is equal to the bid-ask spread. Thus, there is
a threshold above the efficient price at which the informed trader is indifferent between
submitting a BMO or a BLO. Moreover, the equilibrium bid-ask spread turns out to
be equal to the expected decay costs incurred by the informed trader. The next few
sections describe in detail all these results.

3.2 Results

In this section, we show that there exists a homothetic Markov Perfect Equilibrium
(MPE) of the model. We also analyze the optimal strategies of the informed and
uninformed traders, and the resulting expected utility. We then study the resulting
equilibrium limit order book and its evolution in time.
The main difficulty in solving for the equilibrium is the inference problem of the
informed trader. To understand why, consider an informed trader who arrives at date t
vt −vtE
and observes the asset value vt , or equivalently the signal wt = V
. Then, in order to
decide what order to submit, she must be able to estimate for instance the payoff of a
buy limit order (BLO). This is a complex problem, because she must take an average
over all future order flow sequences that lead to the execution of her BLO. It turns out,
however, that this payoff can be described easily if one can compute the information
function g(ρ, w) from Definition A1 in the Appendix. We are only able to estimate g
numerically, but conditional on knowing g, the main formulas in the paper are given in
closed form.
In the next result we verify numerically some properties of the information function
from Definition A1 in the Appendix.

Result 1. For all ρ ∈ (0, 1), the functions g(ρ, w), w − g(ρ, w) and g(ρ, w) − g(ρ, −w)
are strictly increasing in w, and

 ρ(1 + γ) ργ 
max , −2g(ρ, 0) − 2 < α − g(ρ, α). (6)
β β

Also, for the more general information functions in Definition A1, the following results
hold: (i) g(ρ, w, j) decreases in j if w > 0; and (ii) g1 is constant and equal to one.

18
 Theorem 1 shows that there exists a MPE of the model if the conditions stated
in Result 1 are satisfied. We verify these conditions numerically in Internet Appendix
Section 4.

Theorem 1. Suppose the information function g satisfies analytically the conditions

S
from Result 1, and the investor preference parameters satisfy ū ≥ 2
and ν ≥ γ∆. Then,
there exists a homothetic Markov Perfect Equilibrium of the game.

We describe the main properties of the equilibrium in the Corollaries 1–4 below.
Corollary 1 describes the evolution of the efficient price, bid price, and ask price. Corol-
lary 2 describes the initial order submission strategy of the informed trader. Corollary 3
shows that all types of orders are equally likely. Corollary 4 describes the initial strat-
egy of the uninformed traders, and the subsequent equilibrium behavior of all types of
traders in the limit order book.

Corollary 1. In equilibrium, the efficient volatility and the bid-ask spread are constant
and equal, respectively, to the parameters V and S from (3). If the efficient price is vtE ,
the ask price is vtE + S/2, while the bid price is vtE − S/2.
The efficient price changes only when a new order arrives. Let γ ≈ 0.2554 be as in
equation (2). If an order arrives at t, the efficient price changes from vtE to (i) vtE + ∆
if the order is BMO, (ii) vtE + γ∆ if the order is BLO, (iii) vtE − γ∆ if the order is
SLO, and (iv) vtE − ∆ if the order is SMO.

The first part of Corollary 1, that the efficient volatility and the bid-ask spread
are constant over time, follows from the fact that the equilibrium of Theorem 1 is
homothetic. We discuss this issue after Corollary 3.
To get intuition for the second part of Corollary 1, recall that the efficient price is the
expected asset value given the public information (which coincides with the information
of the uninformed traders). The efficient price does not change unless a new order
arrives to the market. A new order affects the efficient price because each type of order
contains private information. To give an example, according to Corollary 2 below, an
informed trader submits a BMO if she observes an extreme asset value, i.e., an asset

19
value vt above vtE + αV , or equivalently a private signal

vt − vtE
wt = (7)
V

above α ≈ 0.6745. By contrast, an informed trader submits a BLO when the signal wt
is positive but moderate, i.e., when the signal wt lies in the interval (0, α). This explains
why a BMO shifts up the efficient price by ∆, while a BLO shifts up the efficient price
only by γ∆ ≈ 0.2554 ∆.
Thus, the key to understanding the equilibrium is the strategy of the informed trader,
which is described in the next result.

Corollary 2. Suppose an informed trader arrives at t ≥ 0, and observes a signal wt =

vt −vtE
V
. Then, she submits a (i) BMO if wt ∈ (α, ∞), (ii) BLO if wt ∈ (0, α), (iii) SLO
if wt ∈ (−α, 0), or (iv) SMO if wt ∈ (−∞, −α). Her expected utility is, respectively,

I S I S
UBLO = + V g(ρ, wt ), USLO = + V g(ρ, −wt ),
2 2 (8)
I S I S
UBMO = − + V wt , USMO = − − V wt .
2 2

To understand this result, suppose the informed trader gets a positive signal wt .
Then, her main choice is between submitting a BMO and a BLO. By submitting a BMO,
she gains from her signal (wt ), but loses half of the bid-ask spread (S/2) because she has
to pay the ask price, which is higher than the efficient price by S/2 (see Corollary 1).
By submitting a BLO instead, we see from equation (8) that the informed trader gains
half of the bid-ask spread, and also benefits from her signal via the information function
g(ρ, w).
The information function increases in w at a lower rate than w itself. Formally, this
follows from Result 1, according to which w − g(ρ, w) is increasing in w. Intuitively, this
is because an informed trader who observes a large signal wt knows that other informed
traders are also likely to receive positive signals in the future, and therefore are more
likely to submit buy orders. This bias towards buy orders therefore pushes up the
efficient price in the future. In other words, the informed trader with a BLO expects to
buy at a higher price in the future while she waits in the book. The stronger her signal,

20
the stronger the bias, and therefore the stronger the relative penalty from submitting
a BLO compared to a BMO. A more detailed discussion of this phenomenon, which is
called the slippage of limit orders, is left for Section 4.
Because the function w − g(ρ, w) is increasing in w, the payoff difference between
BMO and BLO is increasing in w. Therefore, for some threshold α, the informed
trader prefers BMO for wt > α, and BLO for wt ∈ (0, α). Intuitively, with an extreme
signal the informed trader should use a market order, while with a moderate signal
the informed trader should use a limit order. At the threshold w = α (which occurs
with zero probability), the informed trader is indifferent between BMO and BLO. The
threshold α = Φ−1 (3/4) is given by equation (2), and satisfies the property that for a
variable w with the standard normal distribution, the probability that w ∈ (α, ∞) is
equal to the probability that w ∈ (0, α) and is equal to 1/4.

Figure 1: The Order Choice of the Informed Trader. This figure displays (i) the
efficient density, ψtE = N (vtE , V 2 ), which is the density of the asset value vt conditional on
all public information until t, (ii) the four intervals on the horizontal axis that define the
equilibrium order choice of an informed trader after observing vt . The parameter α ≈ 0.6745
is as in equation (2).

vtE − αV vtE vtE + αV

vt
SMO SLO BLO BMO

Figure 1 illustrates the equilibrium order choice of the informed trader. The threshold
between BMO and BLO is given by wt = α, or equivalently by vt = vtE + αV . The

21
normal curve in the figure represents the efficient density, which is the public belief
about the asset value. The four regions under the curve and above the horizontal axis
have area equal to 1/4 each, which reflects the fact that the informed trader submits
each of the four order types with the same ex ante probability. Because the four types
of orders are also equally likely for an uninformed trader,28 , and because there are no
cancellations in equilibrium, it follows that the four types of orders are equally likely in
equilibrium given public information. We state this result in the next corollary.

Corollary 3. Conditional on public information, all order types (BMO, BLO, SLO,
and SMO) are equally likely in equilibrium, with probability 1/4.

We call this equilibrium property dynamic market clearing. It is equivalent to the

following two properties: (i) buy and sell orders are equally likely, and (ii) market and
limit orders are equally likely. It is the second property that is key for the intuition
regarding dynamic market clearing. Suppose for instance that market orders were more
likely than limit orders. Since every market order is executed against a limit order, the
limit order book would become thinner over time, and therefore the equilibrium would
not be homothetic. Thus, dynamic market clearing occurs because the equilibrium in
Theorem 1 is homothetic. In Section 5, we analyze non-homothetic equilibria of the
model, and we see that the dynamic market clearing condition no longer holds.
The next corollary describes the initial order submission decision of the uninformed
traders, as well as their subsequent strategy if they submit a limit order. We only need
to understand the patient uninformed traders, since the impatient ones always submit
market orders. Also, because the informed traders are essentially uninformed after the
initial order choice, the subsequent equilibrium behavior of the informed and uninformed
traders coincides.

Corollary 4. Consider a patient uninformed trader with private valuation u larger in

absolute value than ∆ − S/2. Then, he submits a BLO if he is a natural buyer, and a
28
Indeed, the four types of uninformed traders arrive with equal probability, and the patient natural
buyers submit BLO and the patient natural sellers submit SLO (see Corollary 4), while the impatient
natural buyers submit BMO and the impatient natural sellers submit SMO.

22
SLO if he is a natural seller. In both cases his expected utility is

S
UU = − ∆ + ū. (9)
2

After the initial limit order is submitted, the uninformed trader modifies his order along
with the efficient price, as in Corollary 1. If an informed trader chooses to submit a limit
order, her subsequent behavior mimicks the behavior of an uninformed trader. Traders
in the limit order book modify their orders such that their relative ranks in the bid queue
or the ask queue are always preserved.

The part of Corollary 4 that refers to the uninformed traders is straightforward.

Indeed, a patient natural buyer who submits a BLO gains half of the bid-ask spread
(S/2) and his private valuation (ū), but loses from the adverse selection of the SMO
that eventually executes his order (from Corollary 1, the price impact of a SMO is −∆).
Hence, as long as his private valuation is large enough to make his expected utility
in (9) positive, he optimally submits a BLO. After submitting the initial order, the
uninformed trader simply modifies his order according to the evolution of the efficient
price, because he is risk-neutral and updates his estimate of the asset value according
to the efficient price.
Normally, without additional assumptions one should not expect the equilibrium
limit order book in our model to have a well defined shape. Indeed, trading is for only
one unit, and without any modification cost the exact position of limit orders away from
the bid and ask does not matter. However, the equilibrium shape of the limit order book
can be fixed if we impose an infinitesimal cost of modifying limit orders. Suppose that
when a limit order is executed at the ask, there is an infinitesimal modification cost for
all the remaining limit orders on the ask side (and similarly for the bid side).
The resulting equilibrium limit order book is described in Figure 2. The middle plot
in Figure 2 describes the typical shape of the limit order book just before trading at t. For
simplicity, the efficient price is set at vtE = 0. The right and the left plots, respectively,
describe the effect of a BMO or a BLO on the limit order book. To understand the
assumption about the infinitesimal modification cost, suppose a BMO arrives at date t,
when the limit order book is in the middle plot of Figure 2. Then, the SLO of trader

23
Figure 2: The Effect of Order Flow on the Limit Order Book. This figure displays
the equilibrium shape of the limit order book just before trading at t (middle plot), as well
as the shape of the book at t + 1 after a buy market order BMO (right plot) or a buy limit
order BLO (left plot). For simplicity, the efficient price is set to vtE = 0, so that before trading
E
at t + 1, the efficient price becomes vt+1 = ∆ after BMO, or vt+1 E = γ∆ after BLO. The
parameter γ ≈ 0.2554 is as in equation (2).

BLOt BMOt
LOBt+1 LOBt LOBt+1

(S3) S/2 + 2∆ + γ∆
(S3) S/2 + 2∆ (S3) S/2 + 2∆

(S2) S/2 + ∆ + γ∆
(S2) S/2 + ∆ (S2) S/2 + ∆

(S1) S/2 + γ∆ E
vt+1 =∆
(S1) S/2

E
vt+1 = γ∆ (B1) −S/2 + ∆
vtE = 0

(Bnew) −S/2 + γ∆
(B1) −S/2 (B2) −S/2

(B1) −S/2 − ∆ + γ∆
(B2) −S/2 − ∆ (B3) −S/2 − ∆

(B2) −S/2 − 2∆ + γ∆
(B3) −S/2 − 2∆ (B4) −S/2 − 2∆

(B3) −S/2 − 3∆ + γ∆

S1 is executed, and trader S2 becomes the first in the ask queue. An instant later, S2
should immediately modify his SLO at vtE +S/2+∆, and therefore, with an infinitesimal
modification cost, S2 would prefer to have his order already at that price.

4 Market Quality and Informed Trading

In this section, we consider several measures of market quality and analyze how they
are affected by the informed share, which is the fraction of order flow generated by the
informed traders. As measures of market quality, we consider the information efficiency,

24
as well as three measures of liquidity: the price impact, the bid-ask spread, and the
market resiliency. In the process, we also study the information content of the different
types of orders.

4.1 Information Efficiency

In general, a market is efficient at processing information if pricing errors are small.

In our model, the pricing error is the difference between the fundamental value v and
the efficient price v E , and the standard deviation of this pricing error is the efficient
volatility. But in equilibrium the efficient volatility is constant (see Corollary 1), and it
is equal to the parameter V from (3), which satisfies

V = βρ−1 ∆, (10)

q
2 σv
where ∆ = 1+γ 2
√
λ
, and β and γ are constants defined in (2). Therefore, we propose
the following measure of information efficiency:

1 ρ2
= . (11)
V2 β 2 ∆2

Indeed, when the market is informationally efficient, the efficient volatility is small, and
therefore the proposed measure is large.
Because β and ∆ are independent of ρ, the information efficiency is increasing in
the informed share ρ.29 It follows that information efficiency is increasing with the
informed share. This shows that when there are more informed traders (ρ is large),
the order flow is more informative, hence the market is more efficient at processing
information. An interesting aspect of the increase in information efficiency is that it
arises from the dynamic nature of the equilibrium. In a static equilibrium (see Glosten
and Milgrom 1985), the opposite happens: when there are more informed traders the
adverse selection is larger, and therefore the market is less informationally efficient. This
intuition is discussed in more detail below, after Proposition 1.
The efficient volatility V can be used to estimate in practice the informed share. The
29
The fact that ∆ is independent of ρ is obvious from its formula. The economic interpretation of
this fact, however, is not obvious, and we discuss it in Proposition 1 and the paragraphs that follow it.

25
problem is that it depends on other parameters of the model, such as the fundamental
volatility σv and the total trading activity λ. To remove this dependence, we consider
√
the ratio of the inter-trade volatility (σ = σv / λ) to the efficient volatility (V ),
s
σ 1 + γ2
= ρ ≈ 0.9277 ρ < 1. (12)
V 2β 2

The ratio σ/V provides a clean estimate of the informed share ρ, in the sense that
the ratio does not depend on additional parameters. The inter-trade volatility σ is in
principle observable, as the price variance between trades. The efficient volatility is
not observable directly, but it can be proxied by the dispersion of financial analysts’
estimates. Since (as we show in Section 4.3), the bid-ask spread S is decreasing in the
informed share ρ, a testable implication of equation (12) is that stocks with a lower ratio
of inter-trade volatility to efficient volatility have larger bid-ask spreads.

4.2 Price Impact

We define the price impact of an order as the effect of one additional unit of trading on
the transaction price. Since all trades in our model are for one unit, our marginal price
impact measure is the same as the effect of one unit on the efficient price.30 Because
there are four types of orders, each order type O ∈ {BMO, BLO, SLO, SMO} has a
different price impact, which we denote by ∆O . Corollary 1 implies the following result.

Proposition 1. The price impact ∆O of any order O ∈ {BMO, BLO, SLO, SMO} is

∆BMO = ∆, ∆SMO = −∆, ∆BLO = γ∆, ∆SLO = −γ∆, (13)

q
2 σv σv
where γ ≈ 0.2554 is as in equation (2), and ∆ = 1+γ 2
√
λ
≈ 1.3702 √
λ
is as in
equation (3). In particular, ∆O does not depend on the informed share ρ. Moreover, the
30
Alternatively, given the equilibrium shape of the limit order book (see Figure 2), we can also define
the instantaneous price impact of a multi-unit market order, even though such orders are not part of the
model. Then, as the size of the market order increases, each additional unit trades at a price changed
by ∆. This shows that the two definitions are consistent.

26
 σv2
variance of the price impact is equal to the inter-trade variance σ 2 = λ
, that is,

1 + γ2 2 σ2
∆ = v.


Var ∆O = (14)
2 λ

The reason why all order types have price impact is given by the usual adverse
selection argument. Indeed, when setting the efficient price, the uninformed traders
take into account the information contained in the order flow. For instance, if a BMO
is submitted at t, then with positive probability it comes from an informed trader with
vt −vtE
a large signal wt = V
∈ (α, ∞). Then, the efficient price should increase (by ∆).
Similarly, if a BLO is submitted at t, then it might arrive from an informed trader with
a moderate signal, wt ∈ (0, α). Then, the efficient price should increase as well, although
by a smaller amount (by γ∆).
A surprising implication of Proposition 1 is that the informed share ρ has no effect
on ∆. To give intuition for this result, we note that there are two opposite effects of
the informed share on ∆. Suppose the informed share is small, and a buy market order
arrives. The first effect is the usual adverse selection effect (see for instance Glosten and
Milgrom (1985)): because ρ is small, it is unlikely that the market order comes from
an informed trader. This reduces the adverse selection coming from informed traders,
and therefore decreases the price impact. But there is a second effect, the dynamic
efficiency effect: if the buy market order does come from an informed trader, she must
have observed an asset value far above the efficient price; otherwise, knowing there is
little competition from the other informed traders, she would have submitted a buy
limit order.31 This effect increases the price impact.
Intuitively, the fact that the two effects exactly cancel each other follows from the
equilibrium being homothetic. Indeed, in Internet Appendix Section 7, we show more
generally that in a homothetic equilibrium the change in asset value and the change in
efficient price must have the same variance. In our case, this translates to Var(vt+1 −vt ) =
E
Var(vt+1 − vtE ). But the variance of the asset value change is the inter-trade variance
σ 2 , which does not depend on the informed share, while the variance of the efficient
31
Formally, when ρ is small, the informed trader’s threshold for the choice between BMO and BLO
is large. Indeed, Corollary 2 implies that the threshold signal is wt = α, or equivalently vt = vtE + αV .
But, as discussed in Section 4.1, the efficient volatility V is decreasing in ρ.

27
price change is Var(∆O ), which according to Proposition 1 is a constant multiple of ∆2 .
Therefore, the price impact ∆ is independent of the informed share ρ.
Proposition 1 yields a testable implication of our model, namely that the ratio of
the price impact of a buy market order to the price impact of a buy limit order is

∆BMO 1
= ≈ 3.9152, (15)
∆BLO γ

which is close to 4. Interestingly, Hautsch and Huang (2012, p.515) find empirically
that market orders have a permanent price impact of about four times larger than limit
orders of comparable size.

4.3 Bid-Ask Spread

Another measure of liquidity is the bid-ask spread, which is by definition the difference
between the ask price and the bid price. Corollary 1 implies the equilibrium bid-ask
spread is constant and is equal to the parameter S from equation (3).

Corollary 5. The equilibrium bid-ask spread is constant over time, and is equal to

S = α − g(ρ, α) V. (16)

To get more intuition about the equilibrium bid-ask spread, we explain how the
information function g is interpreted in our model. Consider an informed trader who
v0 −v0E
arrives at t = 0, observes a signal w = V
and submits a BLO (which is not necessarily
optimal). Assuming that subsequently all investors follow their equilibrium strategies,
the informed trader then forms an expectation about the average asset value, based on
all possible future order flow that executes her BLO at a later random time T > 0. The
fact that the BLO is executed at T means that (i) the BLO is the first order in the bid
queue before trading at T , and (ii) a SMO is submitted at T .
To state the next result, we introduce some notation. Let Et be the informed trader’s
expectation conditional on her information set before trading at t, Jt = {w, O1 , . . . , Ot−1 },
and let Ee be the informed trader’s expectation at t = 0 over all future execution se-

28
quences, i.e., over order sequences O1 , . . . , OT that execute the BLO at some T > 0.32

Proposition 2. Consider an informed trader who observes at t = 0 a signal w, and

submits a BLO, which is executed at a random time T > 0. Let ρ ∈ (0, 1) be the informed
share. Then, the information function g satisfies

g(ρ, w) = Ee ET +1 (wT ). (17)

According to Proposition 2, g is the informed trader’s initial expectation of the

signal at execution (wT ) conditional on the execution sequence, including the final SMO
(hence the subscript “T + 1” for the expectation in equation (17)). Then, the difference
w − g(ρ, w) can be interpreted as the signal decay between the initial submission of
the BLO until after its execution. It is therefore a cost that the informed trader faces


when submitting a BLO (relative to submitting a BMO). We call w − g(ρ, w) V the
information decay cost, or simply the decay cost. Corollary 5 implies that the bid-ask
spread S is precisely equal to the decay cost at the threshold signal w = α.


Corollary 6. Let Decay Costw = w − g(ρ, w) V be the information decay cost faced
by an informed trader. Then, the equilibrium bid-ask spread S satisfies

S = Decay Costα . (18)

The intuition for this result is as follows. If the informed trader submits a BMO, she
immediately captures her whole signal (w), but loses half of the bid-ask spread (S/2).
If she submits a BLO instead, she expects the future informed traders to increase the
efficient price by also submitting buy orders, resulting in a decrease of her future signal.
In other words, she expects that by the execution time T the signal wT will decrease
significantly. But this is exactly what the information function g measures. Thus, if
the informed trader submits a BLO, she gains half of the bid-ask spread (S/2), but
captures only part of the signal (g(ρ, w)). Hence, the relative payoff difference between
BMO and BLO is Decay Costw − S. Since at the threshold (w = α) the informed trader
32
The expectation operator Ee is biased, because it is taken only on a subset of the future order flow
sequences (the executable ones). As a result, the law of iterated expectations does not hold. As we
explain below, this bias is caused by the phenomenon of “slippage.”

29
is indifferent between BMO and BLO, it follows that the equilibrium bid-ask spread is
equal to the information decay cost at the threshold.
To get more intuition about the bid-ask spread, we decompose it into two compo-
nents. The first component, called the slippage component, corresponds to the informed
trader’s information decay from the initial submission of the BLO until before its execu-
tion by the final SMO. The second component, called the adverse selection component,
corresponds to the informed trader’s information decay due to the final SMO. To define
these components, we introduce two functions similar to the information function g.

Definition 1. For ρ ∈ (0, 1) and w ∈ R, define the “slippage function” g s (ρ, w) in

the same way as the information function g(ρ, w) from Definition A1 in the Appendix,
ρ
except that the expression µ(Q) = µT +1 − β
is replaced with µ(Q) = µT . Define the
“adverse selection function” as the difference g a = g − g s . Let

Ss = α − g s (ρ, α) V, Sa = S − Ss = g s (ρ, α) − g(ρ, α) V.

(19)

We call S s the “slippage component,” and S a the “adverse selection component.”

The next result provides interpretation of the functions g s and g a .

Proposition 3. In the context of Proposition 2, the slippage function g s and the adverse
selection function g a satisfy

g s (ρ, w) = Ee ET wT ), g a (ρ, w) = Ee ET +1 (wT ) − ET (wT ) .

(20)

Recall that the information function g can be interpreted as the informed trader’s
initial expectation of the signal at execution (wT ) conditional on the execution sequence
including the final SMO. By Proposition 3, the slippage function g s is the same expec-
tation, but conditional on the execution sequence without the final SMO. The difference
is the adverse selection that the informed trader faces at T from the final SMO.
Similar to Corollary 6, the next result shows that both the components of the bid-ask
spread are equal to certain information decay costs. The slippage component is equal to
the information decay cost until the arrival of the final SMO, while the adverse selection
is equal to the information decay cost due to the final SMO.

30


Corollary 7. Define the following cost functions: Slippage Costw = w − g s (ρ, w) V ,
and Adverse Selection Costw = −g a (ρ, w) V . Then, the two components of the bid-ask
spread satisfy

S s = Slippage Costα , S a = Adverse Selection Costα . (21)

Figure 3: Components of the Bid-Ask Spread. This figure plots the bid-ask spread
(S), as well as the slippage component (S s ) and the adverse selection component (S a ). On
the horizontal axis is the informed share ρ = 0.05, 0.10, . . . , 0.95. The bid-as spread and its
components are written in units of the impact parameter ∆.

2.5

S/∆
1.5
Spread / ∆

S a /∆
1

S s /∆
0.5

0
0 0.2 0.4 0.6 0.8 1
Informed Share (ρ)

The next numerical result shows how the bid-ask spread and its components depend
on the informed share ρ.

Result 2. The bid-ask spread S, the slippage component S s , and the adverse selection
component S a are all positive. Moreover, as functions of the informed share ρ, S and
S s are decreasing in ρ, while S a is increasing in ρ.

Figure 3 displays the bid-ask spread and its components against the informed share
ρ. The bid-ask spread and its components are expressed in ∆-units, meaning that we
consider the ratios S/∆, S s /∆, and S a /∆. Using (3) and Definition 1, we compute

S Ss Sa
α − g(ρ, α) βρ−1 , α − g s (ρ, α) βρ−1 , = −g a (ρ, α)βρ−1 . (22)



= =
∆ ∆ ∆

31
The normalization by ∆ does not affect our inferences, because ∆ is independent of ρ
(see Proposition 1). We note that all three terms in (22) contain the factor βρ−1 = V
∆
,
which is strongly decreasing in ρ. This is because, as discussed in Section 4.1, the market
is more informationally efficient when there are more informed traders, which translates
into the efficient volatility V beging smaller. One may thus expect that all three terms
in (22) are decreasing in ρ. Result 2 shows that this is indeed true for the bid-ask spread
S and the slippage component S s , but not for the adverse selection component S a .
The adverse selection component of the bid-ask spread, S a , reflects the fact that the
initial BLO is eventually executed by a SMO coming potentially from a future informed
trader, with superior information. Thus, as expected, adverse selection increases when
there are more future informed traders, that is, S a is increasing in the informed share ρ.
Moreover, S a is close to zero when the ρ approaches zero. This is intuitive, since when
there are few informed traders, there is little adverse selection.33
The slippage component of the bid-ask spread, S s , reflects the phenomenon of slip-
page, which is the signal decay caused by competition with the other informed traders.
When there are more informed traders (the informed share ρ is higher), more informed
traders are likely to arrive in the future, and therefore the rate at which slippage occurs
is higher. The total amount of slippage, however, is multiplied by the efficient volatil-
ity V (recall that signals are normalized by the efficient volatility). Since the efficient
volatility is strongly decreasing in the informed share, the slippage component is actually
decreasing in the informed share, as can be seen in Figure 3.
The bid-ask spread S is the sum of the adverse selection component and the slippage
component. In Figure 3 we see that S is indeed decreasing in the informed share ρ,
although the overall effect is not as strong as the effect of ρ on efficient volatility. When
the informed share increases from ρ = 0.05 to ρ = 0.95, the bid-ask spread decreases
by about 25%. When the informed share ρ is small, the adverse selection component
is close to zero, and therefore most of the bid-ask spread is determined by the slippage
component.
33
In our model, informed traders only observe the asset value once, when they arrive to the market.
We could set up the model so that informed traders continuously observe the fundamental value. In
that case, the adverse selection cost is zero, since all informed traders have the same information. We
conjecture, however, that the slippage cost remains positive, as competition among the informed traders
still imposes a cost on the submission of limit orders.

32
 Note that the slippage of limit orders can be interpreted as an endogenous waiting
cost for the informed trader who decides to submit a limit order. Indeed, even though
the actual waiting cost of a patient investor is zero, the informed investors’ expected
payoff decreases gradually over time because of slippage.34
The bid-ask spread can be used to construct a clean empirical proxy for the informed
share ρ (which does not depend on other parameters such as the fundamental volatility
σv or trading activity λ). Using equation (3), we compute the ratio of inter-trade
√
volatility σ = σv / λ to the bid-ask spread S,
q
1+γ 2
σ ρ 2β 2
= . (23)
S α − g(ρ, α)

By taking this ratio, the dependence of both σ and S on the other parameters of the
model is removed. The inter-trade volatility σ does not depend on ρ, while Result 2
implies that S is decreasing in ρ. Therefore, the ratio σ/S is increasing in ρ.
We summarize the effect of the informed share ρ on our first two liquidity measures:
(i) the equilibrium price impact ∆ is independent on ρ, while (ii) the equilibrium bid-ask
spread S is decreasing in ρ. The reason for this difference is that the price impact is
determined by the uninformed traders, while the bid-ask spread is determined by the
informed traders. Indeed, the price impact is determined by the evolution of the efficient
price over time, which in turn is determined by the uninformed traders. And, because
the traders with limit orders in the book behave identically whether they are informed
or not (see Corollary 4), the exact breakdown between informed and uninformed traders
becomes irrelevant. By contrast, the bid-ask spread is determined by the optimal order
choice of the informed traders, and their order submission strategy can be shown to be
elastic in the bid-ask spread. Thus, the share of informed traders affects the equilibrium
bid-ask spread.
34
Note that a larger informed share implies higher endogenous waiting costs for an informed trader,
holding the mispricing volatility constant. However, the informed trader gradually becomes less in-
formed, and thus her mispricing variance increases over time. Therefore, the exact behavior of the
average waiting costs is ambiguous, and we leave this analysis for future research.

33
4.4 Resiliency

The third dimension of liquidity we consider is market resiliency. Kyle (1985) defines
resiliency as “the speed with which prices recover from a random, uninformative shock.”
Because, as we will see shortly, in our model the speed of price correction is nonlinear
in the size of the shock, we define resiliency as the rate at which a small uninformative
shock is corrected, in the limit when the size of the shock approaches zero.
More formally, we define resiliency from the point of view of an econometrician
who observes a small uninformative shock to the efficient price. Before the shock, the
econometrician has the same belief about the asset value as the uninformed traders (the
efficient density). Suppose at date t the efficient price shifts down by a small positive
amount x, while the efficient volatility remains the same (V ). We do not specify the
cause of this price shift, but one can imagine it as the reaction to the arrival of some
trades that the econometrician knows to be uninformed. Therefore the econometrician
knows that the shock x is uninformative, and expects the mispricing to be corrected.
We then define resiliency as the rate at which the mispricing is corrected, in the limit
when the shift x approaches zero.

Definition 2. Suppose before trading at t, the econometrician believes the asset value


has a normal density vt ∼ N vtE +x, V 2 , or equivalently perceives a mispricing vt −vtE ∼


N x, V 2 , with x ∈ R. Denote by f (x) the average mispricing vt+1 −vt+1
E
after observing
the order at t. The market resiliency coefficient K is defined by

K = 1 − f 0 (0). (24)

Intuitively, if the order at t comes from an informed trader, she is more likely to
observe a positive mispricing, just as the econometrician does. Hence, she is more likely
to submit a buy order, which pushes up the efficient price and reduces the mispricing.
Thus, the econometrician expects the forecast error to become smaller on average, which
for a positive shock x translates into 0 < f (x) < x. If we linearize f near x = 0, we get
f (x) ≈ f 0 (0)x = (1−K)x. Hence, K is the rate at which the mispricing x gets corrected
when x is small, which is indeed our definition of resiliency. Note that the mechanism
behind resiliency is essentially the same as the mechanism behind slippage: the existence

34
of informed traders corrects mispricings over time (resiliency), which generates a cost
for an informed trader who submits a limit order (slippage).

Proposition 4. The market resiliency coefficient equals

2 γφ(0) + (1 − γ)φ(α) 2
K = ρ ≈ 0.8606 ρ2 . (25)
β

Proposition 4 implies that the market is more resilient when the informed share is
larger. This confirms our intuition that a larger share of informed traders results in a
faster correction of pricing errors.
Market resiliency is related to information efficiency. Indeed, the market resiliency
coefficient K from equation (25) is proportional to the information precision measure,
1/V 2 . Thus, in our model, a larger share of informed traders ρ causes the market to be
both more resilient and more informationally efficient.

5 Non-Homothetic Equilibria
We recall from Section 3 that there exists a homothetic equilibrium of the model, mean-
ing that the efficient volatility is constant (and equal to V ). In this section, we allow
the efficient volatility to have a different initial value than the parameter V . This could
happen, for instance, if an uncertainty shock (an unobserved shock to the fundamental
value) suddenly pushes the efficient volatility above V .
In this case, it turns out that we can fully describe the equilibrium once we know the
initial value of the efficient density σ E , or equivalently the initial value of the normalized
efficient volatility,
σE
τ = . (26)
V
We thus define a non-homothetic equilibrium to be an equilibrium for which the initial
normalized efficient volatility τ is different from one.

35
5.1 Properties of Non-Homothetic Equilibria

In Definition A2 in the Appendix we introduce several new parameters: α̃, β̃, γ̃, g̃,
˜ S̃, which are all functions of two variables, ρ and τ . In addition, g̃ is also a
Ṽ , ∆,
function of a third variable, w. Intuitively, we think of ρ as the informed share; of τ
as the normalized efficient volatility, and of w as the signal, or normalized asset value,
(v − v E )/V . However, just like the corresponding parameters from Section 2.1 (without
the tilde above them), the new parameters are defined completely formally.
Proposition 5 shows that for any initial normalized efficient volatility τ there exists
a non-homothetic Markov Perfect Equilibrium (MPE) of the model, as long as the
conditions in Result IA.4 in Internet Appendix Section 5 are satisfied. We verify these
conditions numerically in Internet Appendix Section 6, for τ sufficiently close to one.
The next result also describes several properties of the equilibrium.

Proposition 5. Let τ0 > 0. If the conditions in Result IA.4 are satisfied, there exists a
Markov Perfect Equilibrium of the game in which the initial normalized efficient volatility
is τ0 . In equilibrium, the normalized efficient volatility τt = σtE /V evolves according to


α̃t

2 
0

2 !
− φ α̃τtt


1+γ 2 φ τt
φ τt
2
τt+1 = ρ2 + τt2 − 2ρ2 τt2 1−ρ α̃


+ 1−ρ

, (27)
+ ρ Φ τα̃t − Φ(0)
2


2β 4
+ρ 1−Φ τt 4

˜ t = ∆(ρ,
where α̃t = α̃(ρ, τt ). Let γ̃t = γ̃(ρ, τt ), ∆ ˜ τt ), and S̃t = S̃(ρ, τt ). Then, an order
˜ t if the order is BMO,
arriving at t ≥ 0 changes the efficient price from vtE to (i) vtE + ∆
˜ t if the order is BLO, (iii) v E − γ̃t ∆
(ii) vtE + γ̃t ∆ ˜ t if the order is SLO, and (iv) v E − ∆
˜t
t t

if the order is SMO. At date t, the bid-ask spread is S̃t , the ask price is vtE + S̃t /2, and
the bid price is vtE − S̃t /2.

In equilibrium, the bid-ask spread and the price impact of an order are no longer
constant. The next result, however, provides a linear combination that remains constant
over time. The result involves the parameters S and ∆ from equation (3).

˜ t satisfy
Corollary 8. The equilibrium bid-ask spread S̃t and price impact coefficient ∆

S̃t ˜ t = S − ∆.
−∆ (28)
2 2

36
 Equation (28) is the indifference condition for the uninformed traders. Consider an
uninformed trader who is the first in the bid queue, and suppose his BLO is executed at
˜ t,
date t by a SMO. Then, net of his private valuation, his expected payoff is S̃t /2 − ∆
˜t
where S̃t /2 represents the difference between the efficient price and the bid price, and ∆
represents the adverse selection loss from a potentially informed SMO. If his expected
payoff were not the same at t + 1, then the uninformed traders would have an incentive
to modify their position in the bid queue. The discussion thus far explains why the
˜ t is constant. That the constant is equal to S/2 − ∆ is due to
expected payoff S̃t /2 − ∆
the fact that the equilibrium converges to the homothetic equilibrium of Section 3. We
state this as a numerical result.

Result 3. As t becomes large, the efficient volatility σtE approaches the parameter V ,
˜ t approaches ∆.
the bid-ask spread S̃t approaches S, and the price impact coefficient ∆

5.2 Market Quality in Non-Homothetic Equilibria

We now describe non-homothetic equilibria in more detail, in particular regarding the

market quality measures introduced in Section 4: information efficiency, price impact,
bid-ask spread, market resiliency. To obtain other testable predictions, we also analyze
observable measures such as the limit-to-market impact ratio (the price impact ratio of a
limit order to a market order) and the market-to-limit probability ratio (the probability
ratio of the next order being a market order or a limit order).
First, we analyze a measure that is specific to non-homothetic equilibria: the speed
of convergence to the homothetic equilibrium. Intuitively, this speed is related to the
resiliency of certain market quality measures, such as efficient volatility, bid-ask spread,
or price impact. Indeed, after an uncertainty shock that raises the initial normalized
efficient volatility τ above one, Result 3 above shows that τ (as well as the bid-ask
spread and the price impact) reverts to its homothetic value at a certain rate. It is
then perhaps not surprising that this speed of convergence is closely connected to our
previous measure of market (or price) resiliency, which is the rate at which a mispricing
is corrected.
Figure 4 (left graph) displays the evolution over time of the normalized efficient

37
 Figure 4: Dynamic Information Efficiency. This figure plots the time evolution of two
market quality measures in non-homothetic equilibria. On the horizontal axis is time. Each
curve in each graph corresponds to a value of the informed share (ρ) ranging from 0.05 to 0.95.
The left graph plots the normalized efficient volatility (τt ), which is the efficient volatility (σtE )
divided by its homothetic value (V ); time on the horizontal axis is in logarithmic scale and
is shifted by one, such that time 1 corresponds to t = 0 in the model; the initial normalized
efficient volatility in all cases is τ0 = 2. The right graph plots the bid-ask spread (S̃t ) in
units of the impact parameter ∆; the initial (non-normalized) efficient volatility in all cases is
σ0E = 2∆.

2 2.4
ρ=0.05 ρ=0.40
ρ=0.15 ρ=0.45
1.9 ρ=0.25 ρ=0.50
ρ=0.35 2.3 ρ=0.55
ρ=0.45
1.8 ρ=0.55
Normalized Efficient Volatility ( τ )

ρ=0.65
ρ=0.75 2.2
1.7 ρ=0.85

Bid-Ask Spread /∆
ρ=0.95

1.6 2.1

1.5
2

1.4

1.9
1.3

1.2 1.8

1.1
1.7

1
100 101 102 103 104 0 2 4 6 8 10 12 14 16 18 20
Time ( t + 1 ) Time ( t )

volatility τt according to equation (27), starting from an initial value τ0 = 2. Each

curve in the plot corresponds to an informed share ρ ranging from 0.05 to 0.95. We
observe that in all cases the normalized efficient volatility indeed converges to one, and
furthermore, that the speed of convergence is inversely related to the informed share.
More formally, we define the recovery time to be the number of trading rounds it
takes for the normalized efficient volatility to revert within a neighborhood of one after
a positive or negative shock. In Figure 4, we choose as neighborhood of one the interval
(1 − 10−4 , 1 + 10−4 ). Numerically, the recovery time appears to be linear in the inverse
informed share, 1/ρ2 , regardless of the choice of neighborhood or shock size. We report
this fact as a numerical result.

38
Result 4. The recovery time after a shock to the normalized efficient volatility (τt =
σtE /V ) is linear in the inverse squared informed share (1/ρ2 ).

This result confirms our previous intuition that informed traders make the market
more dynamically efficient. Indeed, when there are more informed traders (the informed
share is higher), a shock that moves the efficient volatility away from its homothetic value
(τ = 1) is followed by a quicker reversal to the homothetic value, and hence it requires
a shorter recovery time. The quicker convergence occurs because orders carry more
information when the informed share is higher, since the probability of each order being
submitted by an informed trader is higher.
The inverse recovery time is thus a measure of information efficiency, and Result 4
shows that this measure is linear in the squared informed share (ρ2 ). In Section 4.1,
another measure of information efficiency is the inverse homothetic efficient variance
(1/V 2 ), which is also linear in the squared informed share.35 The two measures share
the same dynamic efficiency intuition, but the inverse recovery time measure is more
explicit in how dynamic efficiency is achieved.
We now discuss the bid-ask spread S̃ = S̃(ρ, τ ) and the price impact coefficient
˜ = ∆(ρ,
∆ ˜ τ ). From the results of the previous section, we know that the two measures
˜ = S/2 − ∆.36 Thus, they have a similar evolution over time.
are connected by S̃/2 − ∆
Moreover, as we see in Result 5 below, both are increasing functions of τ .
Figure 4 (right graph) displays the evolution over time of the bid-ask spread S̃t if
we start with the same (non-normalized) efficient volatility σ0E = 2∆.37 Intuitively, this
graph shows the effect of an absolute uncertainty shock at t = 0 on the bid-ask spread
S̃, and how that effect depends on the information share ρ. We see that initially the
bid-ask spread S̃ jumps to a higher value when the informed share is higher. This is
because there is more static adverse selection when there are more informed traders.
Over time, however, we see that a higher informed share pushes the bid-ask spread to
a lower value, as the market is dynamically more efficient. Indeed, the bid-ask spread
35
By equation (11), the inverse homothetic efficient variance is proportional to ρ2 .
36 ˜
More formally, Corollary 8 translates into S̃(ρ, τt )/2− ∆(ρ, τt ) = S(ρ)/2−∆(ρ). By setting τ0 = τ ,
˜
it follows that S̃/2 − ∆ = S/2 − ∆ for any τ > 0 and ρ ∈ (0, 1).
37
We only consider the values ρ = 0.40 to ρ = 0.55 because we want τ0 = σ0E /V ∈ (1, 1.5]. We
require τ0 > 1 because we want a positive shock to the normalized efficient volatility, and we require
τ0 ≤ 1.5 because our numerical algorithm has only been made to work for 0.5 ≤ τ0 ≤ 1.5.

39
 S̃ converges over time to the homothetic bid-ask spread S (see Result 3), which is
decreasing in the informed share ρ (see Result 2).
Figure 5: Market Quality in Non-Homothetic Equilibria. This figure plots three
market quality measures in non-homothetic equilibria. On the horizontal axis is the normalized
efficient volatility (τ ), which is the efficient volatility (σ E ) divided by its homothetic value (V ).
Each curve in each graph corresponds to a value of the informed share (ρ) ranging from 0.05
to 0.95. The left graph plots the relative price impact coefficient, which is the price impact
coefficient ∆˜ = ∆(ρ,
˜ τ ) divided by its homothetic value ∆. The middle graph plots the limit-
to-market impact ratio γ̃, which is equal to the price impact of a limit order divided by the
price impact of a market order (γ̃ ∆/ ˜ ∆);
˜ the horizontal line corresponds to the equilibrium
value γ ≈ 0.2554 in the homothetic equilibrium. The right graph plots the market-to-limit
probability ratio PPMOLO
(ρ, τ ), which is the probability the next order is a market order, divided
by the probability that the next order is a limit order. If the numerical procedure does not
yield a unique value, the corresponding point in the plot is omitted.

2 0.5 1.8
ρ=0.05
ρ=0.05 ρ=0.05

Market-to-Limit Probability Ratio (PMO /PLO )

ρ=0.15
ρ=0.15 0.45 ρ=0.15
1.6
Relative Price Impact Coefficient (∆/∆)

ρ=0.25
ρ=0.25 ρ=0.25 ρ=0.35
ρ=0.35 ρ=0.35
˜

0.4 ρ=0.45
Limit-to-Market Impact Ratio

ρ=0.45 ρ=0.45 1.4 ρ=0.55

ρ=0.55 ρ=0.55
ρ=0.65
ρ=0.65 0.35 ρ=0.65
ρ=0.75
1.5 ρ=0.75 ρ=0.75 1.2 ρ=0.85
ρ=0.85 ρ=0.85
ρ=0.95
ρ=0.95 0.3
1
0.25
0.8
0.2
1 0.6
0.15

0.4
0.1

0.05 0.2

0.5 0 0
0.5 1 1.5 0.5 1 1.5 0.5 1 1.5
Normalized Efficient Volatility (τ ) Normalized Efficient Volatility (τ ) Normalized Efficient Volatility (τ )

Because the price impact and bid-ask spread depend on the normalized efficient
volatility τ in the same way, we can focus on either of these measures. We thus consider
˜ or equivalently the relative price impact coefficient which is ∆
the price impact ∆, ˜

divided by its homothetic value, the parameter ∆. Equation (A22) in the Appendix
then implies that the relative price impact coefficient is

˜
∆ β̃(ρ, τ )
= τ. (29)
∆ β

Figure 5 (left graph) plots the dependence of the relative price impact coefficient on

40
both ρ and τ .38 Each curve in the plot corresponds to a value of the informed share
ρ ranging from 0.05 to 0.95. We observe that in all cases the relative price impact is
increasing in τ . We report this fact as a numerical result. Since ∆ does not depend on
˜ Also,
either ρ or τ , the next result is equally true for the price impact coefficient ∆.
equation (8) shows that the same is true for the bid-ask spread S̃.

˜ and the bid-ask spread (S̃) are increasing in

Result 5. The price impact coefficient (∆)
the normalized efficient volatility (τ ).

Intuitively, when the normalized efficient volatility τ is larger, the uninformed traders
have imprecise knowledge about the fundamental value, and therefore the adverse selec-
˜ is larger,
tion is stronger. That implies that the price impact of a buy market order, ∆,
as confirmed by Result 5. The bid-ask spread, S̃, is also larger, to compensate the un-
informed traders for the increase in adverse selection. Formally, the bid-ask spread and
price impact vary with τ in the same way, since equation (28) implies that the difference
˜ is equal to S/2 − ∆, which does not depend on τ .
S̃/2 − ∆
Putting together the previous results, it follows that after a positive shock in the
efficient volatility (or equivalently in τ ), the bid-ask spread S̃ initially increases, to adjust
for the higher value of τ , after which it decreases gradually to its homothetic value S.
˜ Thus, in our model the bid-ask spread
The same effect occurs for the price impact ∆.
and the price impact coefficient both display resiliency, in the sense that they eventually
recover to their homothetic values after a shock in the efficient volatility. We call this
phenomenon liquidity resiliency.
Liquidity resiliency is different from market resiliency. As discussed in Section 4.4,
market resiliency is defined as the recovery of prices after an uninformative shock. In the
context of non-homothetic equilibria, we obtain the following result similar to Proposi-
tion 4.

38
The figure displays results computed with the function g instead of g̃; very similar results are
obtained by using instead the estimated function g̃. The numerical procedure used to solve for the
equilibrium is explained in Internet Appendix Section 6. We impose the strict condition that the
solution to the first equation in (A20) must be unique. When the threshold α̃ is close to zero, this
condition is not satisfied because of estimation errors. This explains why there are missing points in
Figure 5 when ρ is large and τ is small. Intuitively, this occurs because the increase in adverse selection
makes the indifference condition (28) for the uninformed traders harder to satisfy.

41
Proposition 6. The equilibrium market resiliency coefficient K̃ = K̃(ρ, τ ) satisfies

α̃


2ρ2 γ̃φ(0) + (1 − γ̃)φ τ
K̃ = . (30)
τ β̃

Numerically, the market resiliency coefficient K̃ is increasing in the informed share ρ,

and decreasing in the normalized efficient volatility τ . The intuition for why market re-
siliency is increasing in the informed share is the same as in the homothetic equilibrium.
The new result is that market resiliency is decreasing in the normalized efficient volatil-
ity. Intuitively, when the efficient volatility is large, the informed traders become less
aggressive and are more likely to submit limit orders (as we explain below). Therefore,
it takes longer for the price to converge to the fundamental value.
We introduce a new market of market quality, the market-to-limit probability ratio,
which is defined as the probability the next order is a market order, divided by the prob-
ability that the next order is a limit order. In the homothetic equilibrium of Section 3,
this ratio is equal to one since all types of orders are equally likely (see Corollary 3).
In non-homothetic equilibria, the market-to-limit probability ratio varies with both the
informed share and the normalized efficient volatility.

Proposition 7. The market-to-limit probability ratio is equal to

1−ρ
ρ 1 − Φ α̃τ



PMO 4
+
= 1−ρ

. (31)
Φ α̃τ − Φ(0)


PLO 4
+ρ

Figure 5 (right graph) plots the dependence of the market-to-limit probability ratio
on both ρ and τ . Each curve corresponds to a value of the informed share ρ ranging
from 0.05 to 0.95. We observe that in all cases the market-to-limit probability ratio is
decreasing in τ . Intuitively, as explained before, when the normalized efficient volatility
τ is larger, there is an increase in adverse selection for the uninformed traders. This
causes the bid-ask spread, as well as the price impact, to be larger. But the increase
in the bid-ask spread changes the informed traders’ tradeoff between market orders
and limit orders, and makes limit orders more attractive. Thus, the market-to-limit
probability ratio is smaller when the efficient volatility is larger. This result, along with
our previous results regarding the resiliency of the bid-ask spread and the price impact

42
after a efficient volatility shock provide new testable empirical implications.
Figure 5 (middle graph) plots a related measure, the limit-to-market impact ratio γ̃,
˜ to the price impact of a
which is the ratio of the price impact of a buy limit order (γ̃ ∆)
˜ We observe that in all cases the limit-to-market impact ratio is
buy market order (∆).
increasing in τ . The intuition is based on the fact that limit orders are relatively more
likely when τ is larger, which implies that their price impact is also larger. This result
is however dependent on the efficient density being normal, and we therefore consider
it less robust than the previous results.

6 Conclusion
We have presented a model of an order driven market with asymmetric information
in which investors can choose between demanding liquidity with a market order and
providing liquidity with a limit order. Despite the difficulty of the problem, the model
is tractable, and—except for the information function that is computed numerically—
our results are obtained in closed form.
Our main result is that informed trading, as proxied in our model by the informed
share, has an overall positive effect on liquidity, under its three dimensions: tightness
(bid-ask spread), depth (price impact), and resiliency (the speed at which pricing errors
are corrected). In particular, a larger informed share (i) leads to a smaller bid-ask spread,
(ii) generates a stronger market resiliency, yet (iii) does not affect the price impact
of one additional unit of trading. From the perspective of the informed trader, limit
orders have a slippage cost, which measures the erosion in information advantage due to
the competition from future informed traders. Slippage costs represent an endogenous
waiting cost for informed traders, and generate a new component of the bid-ask spread.
We also estimate the information content of order flow. In particular, because in
equilibrium informed traders also use limit orders—whereas in much of the theoretical
literature informed traders only use market orders,—in our model limit orders also have
a nonzero price impact. Quantitatively, we find that the price impact of a limit order is
roughly one fourth of the price impact of a market order.
The results described thus far are true in the context of the homothetic equilibrium,

43
in which the efficient volatility is constant. If an uncertainty shock suddenly increases the
efficient volatility, our results predict that the efficient volatility (as well as the bid-ask
spread and price impact) decrease over time toward the homothetic equilibrium value,
at a speed that is increasing in the informed share. We introduce a new measure, the
market-to-limit ratio, which measures the probability of a trader to submit a market
order relative to a limit order. After an uncertainty shock, the market-to-limit ratio
drops significantly below one, as the increase in the bid-ask spread temporarily convinces
the informed traders to submit more limit orders. The connections among the market-
to-limit ratio with the liquidity measures and the efficient volatility, as well as the
expected evolution of the equilibrium towards the homothetic one, produce new testable
implications of the model.
Our results show that informed trading has an important effect on liquidity, especially
under its resiliency aspect. But estimating market resiliency directly is difficult, since
that would involve having access to information that is not public. Instead, our results
regarding non-homothetic equilibria suggest that we can use liquidity resiliency instead,
which is observable as long as we can identify uncertainty shocks.
Yet another approach is to use rigidities such as stale prices as evidence of low market
resiliency, and study the connection with informed trading. We argue that market
resiliency is inversely related to the price delay measure of Hou and Moskowitz (2005,
in short HM05). HM05 find empirically that firms in which the price responds with a
delay to information command a large return premium.39 Interestingly, HM05 find that
the delay premium has little relation with the PIN measure of Easley, Hvidkjaer, and
O’Hara (2002), which is another measure of informed trading. This suggests that the
informed share in our model may in fact be measuring a different aspect of informed
trading than PIN. Since PIN is based on large imbalances between buyers and sellers,
we postulate that PIN is related to informed trading done by large traders, possibly
corporate insiders. By contrast, the informed share in our model may be more related
to trading done by small informed traders that are not necessarily insiders, and are just
39
Indeed, it is plausible that firms in which prices respond with a delay to information are also firms
for which prices move more slowly toward the fundamental value. It is true that HM05 consider delay
at weekly (or in some robustness checks at daily) frequency, while in our model it is more natural to
think of events as occurring at higher, intra-day frequencies. Then, our identification is correct if delay
at lower frequencies is correlated with delay at higher frequencies.

44
better informed than the public.
Overall, our theoretical model produces a rich set of implications regarding the con-
nection between the activity of informed traders and the level of liquidity. We find
that informed traders have on aggregate a positive effect, by making the market more
efficient and, at the same time, more liquid. A welfare analysis also shows that a larger
number of informed traders (caused for instance by an exogenous decrease in informa-
tion costs) increases aggregate trader welfare. Our model thus provides useful tools to
analyze informed trading, and its connection with liquidity, prices, and welfare.

Appendix
We define formally the information function g that is used throughout the paper. Let
φ( · ; m, s) be the normal density with mean m and standard deviation s. Let φ( · ) =
φ( · ; 0, 1) be the standard normal density, and Φ( · ) its cumulative density. Let 1X be
the indicator function which equals one if X is true and zero if X is false.

Definition A1. Let ρ > 0. For every order O ∈ {BMO, BLO, SLO, SMO}, define δO =
n o
ρ ρ ρ ρ
β
, γ β
, −γ β
, − β
, iO = {(α, ∞), (0, α), (−α, 0), (−∞, −α)}, and jO = {0, +1, 0, −1}.
R
If ψ is a density over R and ΨO = z∈iO ψ(z)dz, define πψ,O and the density fψ,O by

 q 
1−ρ 2
+ ρ1z∈iO ψ(z)φ x; z − δO , ρ 1+γ
R

4 2β 2
dz
1−ρ
πψ,O = 4
+ ρΨO , fψ,O (x) = .
πψ,O
(A1)
Let j0 ≥ 1 be an integer, and Q = (O0 , O1 , . . . , OT ) a sequence of orders, with T ≥ 1.
P 0
We say that Q is an “execution sequence” if j0 + Tt=1 jOt = 0 but j0 + Tt=1 jOt 6= 0 for
P

any 0 ≤ T 0 < T . Then, to every execution sequence Q = (O0 , O1 , . . . , OT ) and density

ψ1 over R, we associate P (Q) = Tt=1 Pt and µ(Q) = µT +1 − βρ , where we recursively
Q

define Pt = πψt ,Ot , ψt+1 = fψt ,Ot , and µt+1 = E(ψt+1 ), for t = 1, . . . , T .
 q 
2
Let ρ ∈ (0, 1) and w ∈ R. Let j0 = 1, and ψ1 = φ · ; w − γ βρ , ρ 1+γ 2β 2
. Then the
“information function” is
X
g(ρ, w) = P (Q)µ(Q), (A2)
Q∈Q

where Q is the set of all execution sequences of the form Q = (BLO, O1 , . . . , OT ).

45
 Let g(ρ, w, j0 ) be defined as above but starting with an arbitrary integer j0 ≥ 1. We
also define the function g1 (ρ, w, j0 ) by taking µ(Q) = 1 in equation (A2); if j0 = 1, we
omit the dependence on j0 and write g1 (ρ, w) = g1 (ρ, w, 1).

Before proving Theorem 1 and discussing Result 1, we explain how investors’ beliefs
are updated after observing the order flow. For an order O = {BMO, BLO, SLO, SMO},
n o
define, respectively, δO = βρ , γ βρ , −γ βρ , − βρ and iO = {(α, ∞), (0, α), (−α, 0), (−∞, −α)}.
Let φ(·; m, s) be the normal density with mean m and standard deviation s, φ(·) the
standard normal density (m = 0, s = 1), and Φ(·) the cumulative normal density.
q 2
Denote the normalized inter-trade volatility by σ̂ = V = ρ 1+γ
σ
2β 2
.

Lemma A1. In the context of Theorem 1, consider a trader who just before trading at
vt −vtE
t believes the signal wt = V
= z has probability density function ψt (z). Then, the
following are true:

(a) The probability of observing O at t is

1−ρ
Z
PO = +ρ ψt (z)dz. (A3)
4 z∈iO

E
vt+1 −vt+1
After seing the order O at t, the posterior density of wt+1 = V
is

1−ρ
R


4
+ ρ1z∈iO ψt (z)φ x; z − δO , σ̂ dz
ψt+1,O (x) = . (A4)
PO

(b) Suppose ψt (·) is not necessarily normal, and has mean µt and standard deviation
σt . Define the normalized price impact δt+1,O to be the change in the expectation
of wt after observing O at t. Then,
R
ρ iO
ψt (z)(z − µt )dz
δt+1,O = E(wt | ψt , O) − E(wt | ψt ) = . (A5)
PO

Denote by µt+1,O and σt+1,O the mean and standard deviation, respectively, of the

 z−µt
2 
posterior density ψt+1,O (x). Let Vt+1,O = P1O ψt (z) 1−ρ
R
4
+ρ1 z∈iO σt
− 1 dz.

46
 Then,

2
µt+1,O = µt − δO + δt+1,O , σt+1,O = σt2 (1 + Vt+1,O ) + σ̂ 2 − δt+1,O
2
, (A6)

E(wt | ψt , O) = µt+1,O + δO = µt + δt+1,O . (A7)

2 2


Let µ̄t+1 = EO µt+1,O and σ̄t+1 = EO σt+1,O , where EO represents the aver-
age over O ∈ {BMO, BLO, SLO, SMO}, with weights PO . Then, EO (δt+1,O ) =
EO (Vt+1,O ) = 0, and

2
µ̄t+1 = µt − EO δO , σ̄t+1 = σt2 + σ̂ 2 − EO δt+1,O
2
. (A8)

LO −µt HO −µt
(c) If ψt (·) = φ(· ; µt , σt ) is normal, let iO = (LO , HO ), `O = σt
, hO = σt
.
Then,

1−ρ

PO = + ρ Φ(hO ) − Φ(`O ) ,
4
2
µt+1,O = µt − δO + δt+1,O , σt+1,O = σt2 (1 + Vt+1,O ) + σ̂ 2 − δt+1,O
2
, (A9)



ρσt φ(`O ) − φ(hO ) ρ `O φ(`O ) − hO φ(hO )
δt+1,O = , Vt+1,O = .
PO PO

If we write µ̄t+1 = f (µt ), then

ρ2
 µ   α + µ   α − µ 
0 t t t
f (µt ) = 1 − 2γφ + (1 − γ) φ +φ . (A10)
βσt σt σt σt

(d) If ψt (·) is the standard normal density, with µt = 0 and σt = 1, then for all O at
t,
1
PO = , δt+1,O = δO , µt+1,O = 0, σ̄t+1 = 1. (A11)
4
Hence, the normalized density ψt has constant volatility.
v −v E
Proof. Conditional on observing wt = t V t = z, the probability of an order O at t is
P Ot = O | wt = z = (1 − ρ) 14 + ρ1z∈iO . Indeed, if the trader at t is uninformed (with

probability 1 − ρ), he submits an order O with equal probability 14 ; if the trader at t is

informed (with probability ρ), she submits an order O if and only if z ∈ IO . Integrating
over z, we obtain PO = 1−ρ
R
4
+ ρ z∈iO ψt (z)dz, which proves (A3).

47
 We now compute the density of the normalized asset value at t + 1 after observing
E
an order O at t. Immediately after t the efficient price moves to vt+1 = v E + ∆O , where
n t o
∆ ρ ∆O ρ ρ ρ ρ
∆O ∈ {∆, γ∆, −γ∆, −∆}. Since V = β , note that δO = V ∈ β , γ β , −γ β , − β .
vt+1 −vt v −(v E +∆ )
If z = wt and δv = write x = wt+1 = t+1 Vt O = δv + z − δO . But δv
V
,
2

has a normal distribution given by N 0, Vσ 2 = N (0, σ̂ 2 ), hence P wt+1 = x | Ot =



O, wt = z = P δv = x − z + δO = φ(x − z + δO ; 0, σ̂) = φ(x; z − δO , σ̂). Compute



also P wt+1 = x, Ot = O | wt = z = P wt+1 = x | Ot = O, wt = z P Ot = O | wt =
z = φ(x; z − δO , σ̂) 1−ρ



4
+ ρ1 z∈i O
. Thus, the posterior density is ψt+1,O (x) = P(wt+1 =
( 1−ρ
R R
+ρ1z∈iO )φ(x;z−δO ,σ̂)ψt (z)dz
x | wt ∼ ψt (z), Ot = O) = R t+1 =x,Ot =O | wt =z)ψt (z)dz
P(w
= 4
.
P(Ot =O | wt =z)ψt (z)dz PO

This proves (A4).

1−ρ

+ρ1
To prove part (b), we start by computing as above P wt = z | Ot = O = 4 PO z∈iO .
R 1−ρ

z( 4 +ρ1z∈iO )ψt (z)dz
Multiplying by z and integrating, we get E wt | ψt , O = PO
, and
R 1−ρ

( 4 +ρ1z∈iO )(z−µt )ψt (z)dz R
by subtracting µt = E wt | ψt we get δt+1,O = PO
. But (z −
R
ρ1z∈iO (z−µt )ψt (z)dz
µt )ψt (z)dz = 0, hence δt+1,O = PO
, which proves (A5).
To compute the mean of ψt+1,O (x), we integrate the formula (A4) over x, and obtain
( 1−ρ
R
+ρ1z∈iO )(z−δO )ψt (z)dz
µt+1,O = 4
PO
. This is similar to the formula we have proved for
δt+1,O , except that µt is replaced by δO . We get µt+1,O = δt+1,O + µt − δO , which proves
the first part of (A6).
For the second part of (A6), note that for any (not necessarily normal) distribution
R
ψ with mean µ and variance σ 2 , (x + a)2 ψ(x)dx = σ 2 + (µ + a)2 . Then,
Z
(x − µt + δO )2 ψt+1,O (x)dx = σt+1,O
2
+ (µt+1,O − µt + δO )2 = σt+1,O
2 2
+ δt+1,O . (A12)

R
We also integrate directly (x − µt + δO )2 ψt+1,O (x)dx by replacing ψt+1,O (x) as in (A4).
R
Using the formula (x − µt + δO )2 φ(x; z − δO , σ̂)dx = (z − µt )2 + σ̂ 2 , we obtain

1−ρ
R

ψt (z) + ρ1z∈iO (z − µt )2 dz
Z
2 2 4
(x − µt + δO ) ψt+1,O (x)dx = σ̂ + . (A13)
PO

2
Putting together (A12) and (A13), we get the desired formula for σt+1,O . Equation (A7)
follows directly from (A5) and (A6). Finally, proving EO (δt+1,O ) = 0 and EO (Vt+1,O ) = 0
is straightforward, which also implies equation (A8).

48
 1−ρ


To prove part (c), first use (A3) to compute PO = 4
+ ρ Φ(hO ) − Φ(`O ) . To
prove the formula for δt+1,O , make the change of variable z 0 = z−µ σt
t
and denote by
R 0 0


ρσt i0 φ(z )z dz ρσt φ(`O )−φ(hO )
i0O = (`O , hO ). Then, δt+1,O = O
PO
= PO
. A similar computation
P
for Vt+1,O finishes the proof of (A9). Finally, µ̄t+1 = f (µt ) = µt − O PO δO = µt −
ρ O Φ(hO ) − Φ(`O ) δO . If we differentiate the endpoints of i0O with respect to µt ,
P

we get − σ1t in all cases, hence f 0 (µt ) = 1 − ρ O φ(hO ) − φ(`O ) − σ1t δO . Using
P


n o
δO ∈ βρ , γ βρ , −γ βρ , − βρ , a straightforward calculation proves (A10).
To prove part (d), we substitute µt = 0 and σt = 1 in the formulas above. We only
prove the results for O = BMO and BLO, the proof for the other order types being
R∞
symmetric. The probability of a BMO is PBMO = 1−ρ 4
+ ρ α φ(z)dz = 1−ρ 4
+ ρ 14 = 41 .
Rα
The probability of a BLO is PBLO = 1−ρ
4
+ ρ 0
φ(z)dz = 1−ρ
4
+ ρ 14 = 14 .
R∞
ρ φ(z)zdz ρφ(α)
The normalized price impact of a BMO is δt+1,BMO = α
PBMO
= 1/4
= βρ =
ρ 0α φ(z)zdz
R
ρ(φ(0)−φ(α))
δBMO . The normalized price impact of a BLO is δt+1,BLO = PBLO
= 1/4
=
φ(0)−φ(α) ρφ(α)
φ(α) 1/4
= γ βρ = δBLO . By symmetry, it follows that δt+1,O = δO for all orders
O ∈ {BMO, BLO, SLO, SMO}.
2 2


We now compute µt+1,O = µt − δO + δt+1,O = µt = 0. Also, σ̄t+1 = EO σt+1,O =
 
) = 41 ( βρ )2 + (γ βρ )2 + (−γ βρ )2 + (− βρ )2 = σ̂ 2 , hence


EO σt2 + σ̂ 2 − δO
2 2
. But EO (δO
2
σ̄t+1 = σt2 + σ̂ 2 − σ̂ 2 = σt2 , from which σ̄t+1 = σt = 1. Thus, the posterior mean is equal
to zero irrespective of the order O at t, while the posterior variance is equal to one on
average. This means that the normalized density N (0, 1) corresponds to a homothetic
equilibrium.

In the next two lemmas, we describe the continuation payoff from submitting a BLO
for either a patient speculator (Lemma A2), or for an uninformed patient natural buyer
(Lemma A3), assuming that all investors follow their equilibrium strategies.
To state the next result, let the execution probability function g1 (ρ, w) be as in
Definition A1. Numerically, we verify that g1 is constant and equal to one (see Result 1),
but for the next lemma we do not need any particular expression for g1 .

Lemma A2. In the context of Theorem 1, consider an informed trader who submits a
vt −vtE
BLO at t after observing the signal wt = V
. Then, if subsequently all traders follow

49
the equilibrium strategies, the continuation payoff of the informed trader is

I S
UBLO = g1 (ρ, wt ) + V g(ρ, wt ). (A14)
2

Proof. We simplify notation and assume that the initial BLO is submitted at t = 0.
Denote by Q the set of all execution sequences Q = O0 = BLO, O1 , . . . , OT −1 , OT =


SMO for the initial BLO. Let Jt be the information set of the informed trader just
before trading at t, which consists of the signal w0 observed at t = 0, and the orders
O0 , . . ., Ot−1 . Let Et be the expectation operator conditional on Jt . At the execution
time T , the bid price is vTE − S/2, therefore

X  
I
UBLO = E0 vT − (vTE − 2s ) | Q P0 (Q)
Q∈Q
(A15)
S X X
= P0 (Q) + I(Q) P0 (Q),
2 Q∈Q Q∈Q

where I(Q) = E0 vT − vTE | Q .
For t = 1, . . . , T + 1, let Pt be the probability of observing the order Ot at t condi-
tional on Jt , ψt the density of wt before trading at t, and µt = Et (wt ) the mean of ψt .
We show that the sequence of probabilities (P1 , . . . , PT ), densities (ψ1 , . . . , ψT +1 ), and
means (µ1 , . . . , µT +1 ) is indeed associated to the execution sequence Q, in the sense of
Definition A1. From equations (A3) and (A4) in Lemma A1, it follows that Pt = πψt ,Ot
and ψt+1 = fψt ,Ot , where π and f are given by equation (A1) in Definition A1.
Next, we show that P0 (Q) coincides with P (Q) = Tt=1 Pt from Definition A1. In-
Q

deed, P0 (Q) = P O1 , . . . , OT | O0 = Tt=1 P Ot | O0 , . . . , Ot−1 = Tt=1 Pt = P (Q).

Q
Q
P
Since by definition g1 (ρ, w0 ) = Q∈Q P (Q), using (A15) we obtain the first half of (A14).
P
We are left to prove that Q∈Q I(Q)P (Q) = V g(ρ, w0 ). First, we show that ψ1 =
2

N w0 − γ βρ , 1+γ
2β 2
, as specified in the definition of g. To see this, note that δO0 = δBLO =
v E −v E
γ βρ . Then, w1 = w0 + v1 V−v0 − 1 V 0 = w0 + v1 V−v0 − γ βρ . Because v1 − v0 ∼ N (0, σ 2 ), we
2 2
have Var v1 V−v0 = Vσ 2 = ρ2 1+γ


2β 2
, where the last equality follows from (12). Since ψ1 is
2

the density of w1 , we obtain indeed ψ1 = N w0 − γ βρ , ρ2 1+γ 2β 2 .
Finally, we show that I(Q) = V µ(Q), where µ(Q) = µT +1 − βρ . The executing order
at T is a SMO, therefore (A7) implies that ET wT | SMOT = µT +1 + δSMO = µT +1 − βρ .

50
Thus, I(Q) = V E0 ET wT | SMOT = V µT +1 − βρ .

For future reference, we note that according to (A7) we have the following decom-
position:
ρ
µT +1 − = µT +1 + δSMOT = µT + δT +1,SMOT , (A16)
β
where, as shown in (A5), δT +1,SMOT is the normalized adverse selection from SMOT .

Lemma A3. In the context of Theorem 1, consider a patient uninformed trader with
private valuation ū, who submits a BLO at t. Then, if subsequently all traders follow
the equilibrium strategies, the continuation payoff of the uninformed trader is

U S
UBLO = ū + − ∆. (A17)
2

Proof. See Internet Appendix Section 1.

Verification of Result 1. See Internet Appendix Section 4.

Proof of Theorem 1. The proof of the theorem depends on the conditions in Result 1
being analytically true. We thus assume that, for all ρ ∈ (0, 1),

g(ρ, w), w − g(ρ, w), and g(ρ, w) − g(ρ, −w) are strictly increasing in w,
 ρ(1 + γ) ργ 
max , −2g(ρ, 0) − 2 < α − g(ρ, α),
β β (A18)
g(ρ, w, j) decreases in j for all w > 0, and

g1 (ρ, w, j) = 1 for all w and j ≥ 1,

where g and g1 are as in Definition A1.

We now define a Markov Perfect Equilibrium (MPE) of the game, by specifying a
strategy profile and a belief system that are compatible with each other. In addition, we
specify a set of state variables that summarizes the payoff-relevant information contained
in each history of the game. As public state variables we choose: the efficient price (vtE )
and the efficient volatility (σtE ), the ask price (at ) and the bid price (bt ), as well as the
bid and ask queues.40 As private state variable we choose the fundamental value (vt ),
40
Because in our model traders can submit orders only for one unit, the limit prices for orders other
than the first ones in the bid and ask queues are not relevant.

51
which is observed by the informed trader at the time of her arrival (t).
Because we want the game to be already in a homothetic equilibrium, we choose
N (0, V 2 ) as the initial efficient density (before trading at t = 0). Moreover, the ask
price is S/2, the bid price is −S/2, with S as in (3), while the initial limit order book
has countably many limit orders on each side (see the middle plot in Figure 2).
To define the strategy profile S, we first describe the action of a new trader who
arrives at t. Then, we describe the reaction of the other traders remaining in the
limit order book to the new arrival at t. Finally, in Internet Appendix Section 1 we
describe the reaction of the existing traders to any out-of-equilibrium deviation that
might occur from either the new trader or an existing trader. Recall that impatient
traders are assumed to automatically submit market orders. We therefore describe
only the strategies of patient traders, who can be informed (with private valuation
0), uninformed buyers (with private valuation ū), or uninformed sellers (with private
valuation −ū). The strategy profile S is then given by the following set of rules:

(a) The uninformed buyer arriving at t submits a BLO at the price (vtE + γ∆) − S/2.

(b) The uninformed buyer arriving at t submits a SLO at the price (vtE − γ∆) + S/2.

(c) The informed trader who observes an asset value vt when she arrives at t submits an
vt −vtE
order O ∈ {BMO, BLO, SLO, SMO} whenever her signal V
lies, respectively,
in the interval {(α, ∞), (0, α), (−α, 0), (−∞, −α)}.

(d) After the initial order submission, all traders behave as described in (e) and (f).

(e) If a BMO is submitted at t, then an instant later the efficient price is updated to
vtE + ∆, the ask price to vtE + ∆ + S/2, and the bid price to vtE + ∆ − S/2, and all
other limit traders shift their orders by ∆ such that the relative ranks in the ask
and bid queues are preserved. After that, no other trader moves until t + 1.

The reaction to a SMO at t is symmetric to the reaction to a BMO.

(f) If a BLO is submitted at t, then an instant later the efficient price is updated to
vtE + γ∆, the ask price to vtE + γ∆ + S/2, and the bid price to vtE + γ∆ − S/2, and

52
 all other limit traders shift their orders by γ∆ such that the relative ranks in the
ask and bid queues are preserved. After that, no other trader moves until t + 1.

The reaction to a SLO at t is symmetric to the reaction to a BLO.

For brevity, we leave the description of out-of-equilibrium moves S(g) and S(h) to
Internet Appendix Section 1.
The belief system is described by the following rules: At t = 0, the uninformed
investors perceive the asset value distributed according to N (0, V 2 ). Subsequently, the
uninformed investors’ belief about the asset value (the efficient density) is updated using
the approximate Bayes’ rule described in Section 2. At the time of arrival to the market,
the informed investor observes the asset value and can compute the average payoff of a
limit order based on updating her belief according to the exact Bayes’ rule. After the
arrival, however, the informed trader cannot update her belief, and becomes essentially
uninformed. In the limit order book at t = 0, all traders are uninformed with probability
one. At t ≥ 0, each new trader is believed to be informed with probability ρ by the
other traders. Subsequently, traders’ beliefs about the other investors’ types are updated
according to the Bayes’ rule.
Because the strategy profile S defined above depends only on the current value of
the state variables, the strategies are indeed Markov. We now show that the strategy
of each type of investor is a best response to the other investors’ strategies.

Uninformed Traders

We consider a patient natural buyer (with private valuation ū and zero waiting costs).
We need to show that the strategy specified by S is optimal for this trader. Because the
proof is straightforward but tedious, we present only the intuition behind the results,
and leave the complete proof of this statement to Internet Appendix Section 1.
Intuitively, it is clear that the patient natural buyer chooses a buy order, since with
a sell order he would lose the private valuation ū. Hence, the main choice is between a
BMO, a BLO, and NO (no order). Recall that a simplifying assumption in Section 2 is
that the uninformed trader starts with a prior belief at t such that after submitting his
order his posterior belief coincides with the efficient density. In the proof of Lemma A3

53


we compute that his prior belief is vt ∼ N vtE + γ∆, V 2 − σ 2 (see equation (IA.2) in
Internet Appendix Section 1).
Then, Lemma A3 shows that the trader’s continuation payoff from submitting a BLO
U U
is UBLO = S/2 − ∆ + ū. If instead he submits a BMO, he gets UBMO = Et (vt ) − at + ū =
(vtE + γ∆) − (vtE + S/2) + ū = ū + γ∆ − S/2. Finally, if he submits no order, he gets
by convention zero.
First, we note that BMO is preferred to NO, since ū ≥ S/2. To compare BMO with
BLO, note that condition (A18) implies α − g(ρ, α) > βρ (1 + γ), which if we multiply by
V = βρ−1 ∆ implies S > ∆(1 + γ) and hence UBLO
U U
> UBMO . Thus, the patient natural
buyer optimally submits a BLO. The rest of the proof is in Internet Appendix Section 1.

Informed Traders

We prove that the strategy of an informed trader is as specified in S(c), S(e), S(f).
Consider a (patient) informed trader who arrives at t and observes the asset value vt ,
vt −vtE
or equivalently the signal wt = V
. The informed trader has the option to submit
either (i) BMO, (ii) SMO, (iii) NO (no order), (iv) BLO at b∗ = (vtE + γ∆) − S/2, and
later follow S, (v) SLO at a∗ = (v E − γ∆) + S/2, and later follow S. We show that the
informed trader submits O ∈ {SMO, SLO, BLO, BMO} whenever wt lies, respectively,
in the interval {(−∞, −α), (−α, 0), (0, α), (α, ∞)}; for this, we show that option (iii) is
eliminated by a penalty for not trading that satisfies ν ≥ γ∆. Then, if wt > 0, we
show that option (iv) is less profitable if the BLO is submitted at a different price than
b∗ ; symmetrically, if wt < 0, we show that option (v) is less profitable if the SLO is
submitted at a different price than a∗ . After submitting (iv) or (v), the informed trader
behaves in the same way as the uninformed buyer.
Let UOI be the continuation payoff from submitting O and later following S. As in the
case of the uninformed buyer, we assume that the current limit order book has the ask
I
UO
price at = vtE + S/2, and the bid price bt = vtE − S/2. Let ÛOI = V
be the normalized
S ν
payoff from O; Ŝ = V
the normalized spread parameter; and ν̂ = V
the normalized
I
commitment parameter, which is a penalty for non-trading. From Lemma A2, ÛBLO =
Ŝ I Ŝ
2 1
g (ρ, wt )+g(ρ, wt ). But, by condition (A18), g1 (ρ, wt ) = 1, hence ÛBLO = 2
+g(ρ, wt ).

54
Putting together all the formulas, we get:

I Ŝ I Ŝ I
ÛBMO = wt − , ÛSMO = − − wt , ÛNO = −ν̂,
2 2 (A19)
I Ŝ I Ŝ
ÛBLO = + g(ρ, wt ), ÛSLO = + g(ρ, −wt ).
2 2

Denote by A(w) = w − g(ρ, w), B(w) = w − g(ρ, −w), D(w) = g(ρ, w) − g(ρ, −w); and
I I
note that B(w) = A(w)+D(w). With these notations, we get ÛBMO − ÛBLO = A(wt )− Ŝ,
I I I I I I
ÛBMO − ÛSLO = B(wt ) − Ŝ, ÛBLO − ÛSLO = D(wt ), ÛBLO − ÛSMO = Ŝ − B(−wt ),
I I
ÛSLO − ÛSMO = Ŝ − A(−wt ). From (A18), it follows that that A, D, and B = A + D are
strictly increasing in w, therefore all the payoff differences above are strictly increasing
in wt . Note that by the definition of S, we have A(α) = α − g(ρ, α) = Ŝ, therefore
BMO is preferred to BLO if and only if wt > α. Similarly, SMO is preferred to SLO if
and only if wt < −α. Also, D(0) = 0, therefore BLO is preferred to SLO if and only if
wt > 0. Because all the payoff differences are strictly increasing in wt , a straightforward
analysis shows that indeed the informed trader prefers O ∈ {BMO, BLO, SLO, SMO}
whenever wt lies, respectively, in the interval {(α, ∞), (0, α), (−α, 0), (−∞, −α)}.
Next, we make sure that NO (“No Order”) is never optimal. For that, we use
equation (A19) to compute the minimum payoff for each type of order. According to
condition (A18), g is strictly increasing in w, therefore ÛOI is increasing in wt for BMO
and BLO, and decreasing in wt for SMO and SLO. Thus, it is sufficient to verify that
I
ÛBLO ≥ −ν̂ when wt = 0. Since by assumption ν ≥ γ∆, the formula V = βρ−1 ∆ implies
ργ
ν̂ ≥ β
. Hence, it is sufficient to verify that Ŝ
2
+ g(ρ, 0) ≥ − ργ
β
. But this follows from
condition (A18), which implies that α − g(ρ, α) > −2g(ρ, 0) − 2 ργ
β
.
We now show that the continuation payoff for the informed trader from submitting
his BLO at the equilibrium price b∗ is higher than the payoff obtained by choosing BLO
at either b > b∗ or b < b∗ . We first rule out BLO at b > b∗ . Based on the out-of-
equilibrium reaction S(h) described in Internet Appendix Section 1, overshooting a bid
is interepreted as coming with probability one from an informed trader with a positive
signal. This leads to a positive shift in the efficient price vtE and therefore to a negative
vt −vtE
shift in the informed trader’s signal wt = V
. Condition (A18) then implies that
g(ρ, wt ) strictly decreases, and along with it the informed trader’s expected payoff.

55
 We also rule out BLO at b < b∗ . Based on the out-of-equilibrium reaction S(h)
described in Internet Appendix Section 1, this deviation does not bring any new infor-
mation about the transgressor’s type, but prompts another trader in the bid queue to
immediately modify his BLO at b∗ . The informed trader thus loses his first rank in the
bid queue, which according to Lemma A2 generates a normalized continuation payoff
Ŝ
of 2 1
g (ρ, wt , j) + g(ρ, wt , j), where j > 1 is the informed trader’s new rank in the bid
queue.41 By condition (A18), g1 (ρ, wt , j) = 1 and g(ρ, wt , j) is decreasing in j, which
Ŝ I
implies that the informed traders gets a smaller payoff than 2
+ g(ρ, wt , j = 1) = ÛBLO .
Hence, the informed trader reduces his payoff by deviating from b = b∗ .
Finally, after the initial order choice the strategy of the informed trader is the same
as for the uninformed trader, since they now have the same information set.

Proof of Corollary 1. The corollary follows directly from the description of the equi-
librium strategy profile S, and in particular from S(e) and S(f).

Proof of Corollary 2. This corollary follows directly from the description of the equi-
librium strategy profile S, and in particular from S(c). The formula for the expected
utility of the informed trader follows from equation (A19).

Proof of Corollary 3. As proved in Theorem 1, the efficient density at t is N (vtE , V 2 ),

vt −vtE
which implies that the normalized efficient density (the density of the signal wt = V
)
is standard normal. Then, part (d) of Lemma A1) shows that all orders have probability
equal to 1/4.

Proof of Corollary 4. This corollary follows directly from the description of the equi-
librium strategy profile S, and in particular from S(a) and S(b). The formula for the
expected utility of the uninformed trader follows from Lemma A3.

Proof of Proposition 1. From Corollary 1, any order O ∈ {BMO, BLO, SLO, SMO}
moves the efficient price vtE by ∆O ∈ {∆, γ∆, −γ∆, −∆}, respectively. Because each
type of order occurs with probability 41 , and the efficient price moves by {∆, γ∆, −γ∆, −∆},
E 1+γ 2 2
it is simple to show that the variance of vt+1 − vtE is indeed equal to 2
∆.
41
By condition (A18), g1 = 1.

56
Proof of Corollary 5. By Corollary 1, if the efficient price is v E , at any time the ask
price is v E + S/2, and the bid price is v E − S/2. This implies that the bid-ask spread is


equal to the parameter S = α − g(ρ, α) V from (3), and is therefore constant.

Proof of Proposition 2. The proposition follows from the proof of Lemma A2 in the
Appendix.


Proof of Corollary 6. By equation (3), S = α − g(ρ, α) V = Decay Costα .

Proof of Proposition 3. Recall that the slippage function g s follows Definition A1,
except that µ(Q) = µT instead of µ(Q) = µT +1 − βρ for the information function g. This
proves the formula g s (ρ, w) = Ee ET wT ). The adverse selection function g − g s = g a
therefore follows Definition A1, except that µ(Q) = (µT +1 − βρ )−µT . But equation (A16)
implies µT +1 − βρ = µT +δT +1,SMOT . Hence, g a is defined using µ(Q) = δT +1,SMOT , which is
the price impact of the SMO at execution time T . But this is equal to ET +1 (wT )−ET (wT ),
which proves that indeed g a (ρ, w) = Ee ET +1 (wT ) − ET (wT ) .

Proof of Corollary 7. By equation (19), the slippage component satisfies S s = α −

g s (ρ, α) V = Slippage Costα . Also, the adverse selection component satisfies S a =


g s (ρ, α) − g(ρ, α) V = −g a (ρ, w) V = Adverse Selection Costα .

Verification of Result 2. See Internet Appendix 4.

Proof of Proposition 4. With the notation of Lemma A1 in this Appendix, consider

vt −vtE
a trader that perceives the signal wt = V
distributed according to the normalized
density prior, N (µt , σt2 ) = N (0, 1). Denote by µt+1 = f (µt ) the average posterior mean.
Then, by setting µt = 0 and σt = 1 in equation (A10), we obtain the desired formula
for the resiliency coefficient K.

˜ and S̃ that are used in

We now define formally the parameters α̃, β̃, γ̃, g̃, Ṽ , ∆
Section 5 of the paper. Recall that φ( · ) is the standard normal density, and Φ( · ) is its
cumulative density.

57
Definition A2. Let ρ > 0 and τ > 0. We define the functions α̃ = α̃(ρ, τ ), β̃ = β̃(ρ, τ ),
γ̃ = γ̃(ρ, τ ), and g̃ = g̃(ρ, w, τ ) by the implicit equations:

ρ τ φ α̃τ


ρ
α̃ − g̃(ρ, α̃, τ ) − 2 1−ρ α̃


= α − g(ρ, α) − 2 ,
4
+ ρ 1 − Φ τ
β
1−ρ α̃

1−ρ (A20)
φ(0) − φ α̃τ + ρ 1 − Φ α̃τ





4
+ ρ 1 − Φ τ 4
β̃ = , γ̃ = 1−ρ

,
φ α̃τ φ α̃τ


α̃


4
+ ρ Φ τ
− Φ(0)

while g̃(ρ, w, τ ) is defined as in Definition A1 in this Appendix, by adding a tilde over

the corresponding parameters,42 and letting τ evolve according to τ0 = τ and

α̃t

2 
0

2 !
− φ α̃τtt


1+γ 2 φ τt
φ τt
2
τt+1 = ρ2 + τt2 − 2ρ2 τt2 1−ρ α̃


+ 1−ρ

, (A21)
+ ρ Φ τα̃t − Φ(0)
2


2β 4
+ρ 1−Φ τt 4

˜ =
where α̃t = α̃(ρ, τt ). Also, if V = V (ρ) as in equation (3), define Ṽ = Ṽ (ρ, τ ), ∆
˜ τ ), and S̃ = S̃(ρ, τ ) by
∆(ρ,

˜ = ρ τ V,


Ṽ = τ V, ∆ S̃ = α̃ − g̃(ρ, α̃, τ ) V. (A22)
β̃

Finally, the proofs of Propositions 5–7 and Corollary 8, as well as the verification of
Results 4–5 are left to the Internet Appendix.

References
[1] Amihud, Y. (2002): “Illiquidity and Stock Returns: Cross-Section and Time-Series Effects,”
Journal of Financial Markets, 5, 31–56.
[2] Amihud, Y., and H. Mendelson (1986): “Asset Pricing and the Bid-Ask Spread,” Journal of
Financial Economics, 17, 223–249.
[3] Anand, A., S. Chakravarty, and T. Martell (2005): “Empirical Evidence on the Evolution
of Liquidity: Choice of Market versus Limit Orders by Informed and Uninformed Traders,” Journal
of Financial Markets, 8, 289–309.
[4] Back, K., and S. Baruch (2007): “Working Orders in Limit Order Markets and Floor Ex-
changes,” Journal of Finance, 62, 1589–1621.
[5] Bagehot, W. (1971): “The Only Game in Town,” Financial Analysts Journal, 27, 12–14 and 22.
[6] Bergin, J., and B. MacLeod (1993): “Continuous Time Repeated Games,” International
Economic Review, 34, 21–37.
q
2
42
The only case when a tilde is not added over a parameter is when the term ρ 1+γ2β 2 occurs. This
term is equal to Vσ , the normalized inter-trade variance, where σ is the volatility of the change in
fundamental value between trades.

58
 [7] Biais, B., P. Hillion, and C. Spatt (1995): “An Empirical Analysis of the Limit Order Book
and the Order Flow in the Paris Bourse,” Journal of Finance, 50, 1655–1689.
[8] Biais, B., J. Hombert, and P.-O. Weill (2013): “Equilibrium Pricing and Trading Volume
under Preference Uncertainty,” Review of Economic Studies, forthcoming.
[9] Bloomfield, R., M. O’Hara, and G. Saar (2005): “The ’Make or Take’ Decision in an
Electronic Market: Evidence on the Evolution of Liquidity,” Journal of Financial Economics, 75,
165–199.
[10] Brennan, M., and A. Subrahmanyam (1996): “Market Microstructure and Asset Pricing: On
the Compensation for Illiquidity in Stock Returns,” Journal of Financial Economics, 41, 441–464.
[11] Chakravarty, S., and C. Holden (1995): “An Integrated Model of Market and Limit Orders,”
Journal of Financial Intermediation, 4, 213–241.
[12] Cohen, K., S. Maier, R. Schwartz, and D. Whitcomb (1981): “Transaction Costs, Order
Placement Strategy, and Existence of the Bid-Ask Spread,” Journal of Political Economy, 89,
287–305.
[13] Collin-Dufresne, P., and V. Fos (2013), “Do Prices Reveal the Presence of Informed Trad-
ing,” Working Paper.
[14] Easley, D., and M. O’Hara (1992): “Time and the process of Security Price Adjustment,”
Journal of Finance, 47, 577–605.
[15] Easley, D., and M. O’Hara (2004): “Information and the Cost of Capital,” Journal of Finance,
59, 1553–1583.
[16] Easley, D., S. Hvidkjaer, and M. O’Hara (2002): “Is Information Risk a Determinant of
Asset Returns?” Journal of Finance, 57, 2185–2221.
[17] Easley, D., N. Kiefer, M. O’Hara, and J. Paperman (1996): “Liquidity, Information, and
Infrequently Traded Stocks” Journal of Finance, 51, 1405–1436.
[18] Foucault, T. (1999): “Order Flow Composition and Trading Costs in a Dynamic Limit Order
Market,” Journal of Financial Markets, 2, 99–134.
[19] Foucault, T., O. Kadan, and E. Kandel (2005): “Limit Order Book as a Market for Liquid-
ity,” Review of Financial Studies, 18, 1171–1217.
[20] Fudenberg, D., and J. Tirole (1991): Game Theory, MIT Press.
[21] Glosten, L. (1994): “Is the Electronic Open Limit Order Book Inevitable?,” Journal of Finance,
49, 1127–1161.
[22] Glosten, L., and P. Milgrom (1985): “Bid, Ask and Transaction Prices in a Specialist Market
with Heterogeneously Informed Traders,” Journal of Financial Economics, 14, 71–100.
[23] Goettler, R., C. Parlour, and U. Rajan (2005): “Equilibrium in a Dynamic Limit Order
Market,” Journal of Finance, 60, 2149–2192.
[24] Goettler, R., C. Parlour, and U. Rajan (2009): “Informed Traders and Limit Order Mar-
kets,” Journal of Financial Economics, 93, 67–87.
[25] Griffiths, M., B. Smith, A. Turnbull, and R. White (2000): “The Costs and Determinants
of Order Aggressiveness,” Journal of Financial Economics, 56, 65–88.
[26] Grinstead, C. and L. Snell (2003): Introduction to Probability, American Mathematical Soci-
ety, 2nd Edition.
[27] Harris, L. (1998): “Optimal Dynamic Order Submission Strategies in Some Stylized Trading
Problems,” Financial Markets, Institutions and Instruments, Vol. 7, No. 2.
[28] Harris, L., and J. Hasbrouck (1996): “Market versus Limit Orders: the Superdot Evidence
on Order Submission Strategy,” Journal of and Financial and Quantitative Analysis, 31, 213–231.

59
[29] Hasbrouck (1993): “Assessing the Quality of a Security Market: A New Approach to
Transaction-Cost Measurement,” Review of Financial Studies, 6, 191–212.
[30] Hautsch, N., and R. Huang (2012): “The Market Impact of a Limit Order,” Journal of
Economic Dynamics and Control, 36, 501-522.
[31] Hollifield, B., R. Miller, and P. Sandås (2004): “Empirical Analysis of Limit Order Mar-
kets,” Review of Economic Studies, 71, 1027–1063.
[32] Hou, K., and T. Moskowitz (2005): “Market Frictions, Price Delay, and the Cross-Section of
Expected Returns,” Review of Financial Studies, 18, 981–1020.
[33] Huang, R., and H. Stoll (1997): “The Components of the Bid-Ask Spread: A General Ap-
proach,” Review of Financial Studies, 10, 995–1034.
[34] Jain, P. (2005): “Financial Market Design and the Equity Premium: Electronic versus Floor
Trading,” Journal of Finance, 60, 2955–2985.
[35] Kaniel, R., and H. Liu (2006): “So What Orders Do Informed Traders Use?” Journal of
Business, 79, 1867–1913.
[36] Kyle, Albert (1985): “Continuous Auctions and Insider Trading,” Econometrica, 53, 1315–1335.
[37] Large, J. (2009): “A Market-Clearing Role for Inefficiency on a Limit Order Book,” Journal of
Financial Economics, 56, 65–88.
[38] Latza, T., and R. Payne (2011): “Forecasting Returns and Trading Equities Intra-Day Using
Limit Order and Market Order Flows,” Working Paper.
[39] Linnainmaa, J. (2010): “Do Limit Order Alter Inferences about Investor Performance and Be-
havior?,” Journal of Finance, 65, 1473-1506.
[40] Linnainmaa, J., and I. Roşu (2011): “Weather and Time-Series Determinants of Liquidity in
a Limit Order Market,” Working Paper.
[41] Menkhoff, L., C. Osler, and M. Schmeling (2010): “Limit-Order Submissions Strategies
under Asymmetric Information,” Journal of Banking and Finance, 34, 2665–2677.
[42] O’Hara, M. (1995): Market Microstructure Theory, Blackwell.
[43] Pagnotta, E. (2010): “Information and Liquidity Trading at Optimal Frequencies,” Working
Paper.
[44] Parlour, C. (1998): “Price Dynamics in Limit Order Markets,” Review of Financial Studies, 11,
789–816.
[45] Parlour, C., and D. Seppi (2008): “Limit Order Markets: A Survey,” Chapter 3 in Handbook
of Financial Intermediation and Banking, edited by A.W.A. Boot and A.V. Thakor.
[46] Pástor, Ľ., and R. Stambaugh (2003): “Liquidity Risk and Expected Stock Returns?” Journal
of Political Economy, 111, 642–685.
[47] Rourke, T. (2009): “The Information Content of the Supply and Demand for Liquidity in Stock
and Option Markets,” Working Paper.
[48] Roşu, I. (2009): “A Dynamic Model of the Limit Order Book,” Review of Financial Studies, 22,
4601–4641.
[49] Sandås, P. (2001): “Adverse Selection and Competitive Market Making: Empirical Evidence
from a Limit Order Market,” Review of Financial Studies, 14, 705–734.
[50] Seppi, D. (1997): “Liquidity Provision with Limit Orders and a Strategic Specialist,” Review of
Financial Studies, 10, 103–150.
[51] Wang, J. (1993) “A Model of Intertemporal Asset Prices Under Asymmetric Information,” Review
of Economic Studies, 60, 249–282.

60