Financial Intermediaries and Liquidity Creation

Author(s): Gary Gorton and George Pennacchi

Source: The Journal of Finance , Mar., 1990, Vol. 45, No. 1 (Mar., 1990), pp. 49-71
Published by: Wiley for the American Finance Association

Stable URL: http://www.jstor.com/stable/2328809

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

and Wiley are collaborating with JSTOR to digitize, preserve and extend access to The Journal
of Finance

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
THE JOURNAL OF FINANCE * VOL. XLV, NO. 1 * MARCH 1990

Financial Intermediaries and Liquidity Creation

GARY GORTON and GEORGE PENNACCHI*

ABSTRACT

Trading losses associated with information asymmetries can be mitigated by designing

securities which split the cash flows of underlying assets. These securities, which can
arise endogenously, have values that do not depend on the information known only to
informed agents. Bank debt (deposits) is an example of this type of liquid security which
protect relatively uninformed agents, and we provide a rationale for deposit insurance
in this content. High-grade corporate debt and government bonds are other examples,
implying that a money market mutual fund-based payments system may be an alter-
native to one based on insured bank deposits.

A WIDELY HELD VIEW is that the investor of modest means is at a disadvantage

relative to large investors. This popular perception, dating from at least the early
19th century, has it that the small, unsophisticated investor-"the farmer,
mechanic, and the laborer"-is least equipped to acquire information and is most
often victimized by having to trade with better informed agents. U.S. history is
repeatedly marked by incidents of real or imagined insider shenanigans and
resulting popular initiatives against the "money trusts" and the "robber barons."
This view is responsible for many institutions, e.g., the SEC antitrust legislation,
and various forms of taxation. This argument has also influenced bank regulation
where it has been used to justify government provision of deposit insurance as a
matter of public policy.
The notion that informed agents can exploit uninformed agents has received
some support from Kyle (1985) and Grinblatt and Ross (1985). They show that
insiders can systematically benefit at the expense of uninformed traders when
prices are not fully revealing. However, in these models the uninformed traders,
called noise traders, are nonoptimizing agents; they simply trade and lose money.
If informed agents can, somehow, systematically take advantage of uninformed
agents, then it seems clear that the uninformed agents would be motivated to
respond, possibly creating alternative mechanisms. In this essay we investigate
whether financial institutions and security contracts will endogenously arise as
a response to problems faced by uniformed investors with a need to transact. In

* Both authors from Finance Department, The Wharton School, University of Pennsylvania. A
previous version of this paper was entitled "Transactions Contracts." The comments and suggestions
of Mark Flannery, Jeff Lacker, Chris James, Dick Jefferis, Bruce Smith, Chester Spatt, an anonymous
referee, members of the University of Pennsylvania Macro Lunch Group, especially Randy Wright
and Henning Bohn, and participants in the 1988 NBER Summer Institute, the 1988 Garn Institute
Conference on Federal Deposit Insurance and the Structure of Financial Markets, the 1988 Winter
Econometric Society Meetings, and the Federal Reserve Bank of Richmond were greatly appreciated.
The first author thanks the NSF for financial support through #SES-8618130. Errors remain the
authors'.

49

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
50 The Journal of Finance

particular, we ask whether there are a variety of solutions and whether govern-
ment intervention might be a necessary feature of any of them.
We first consider an environment that is similar in spirit to the above
traditional notion that investors might need to trade in markets where better
informed agents are present. The uninformed agents in our model have uncertain
consumption preferences but are optimizing agents. Like the previous research,
we show that the informed agents may exploit the uninformed, even though here
they are optimizing. However, this result holds only when certain contractual
responses by the uninformed agents are precluded. We go on to consider how the
uniformed agents would respond in order to protect themselves from losses to
the insiders.
The central idea of the paper is that trading losses associated with information
asymmetries can be mitigated by designing securities which split the cash flows
of underlying assets. These new securities have the characteristic that they can
be valued independently of the possible information known only by the informed.
By using these securities for transactions purposes, the uninformed can protect
themselves. While our focus is on trading contexts, Myers and Majluf (1984)
have considered a related problem in corporate finance. When firm managers
have inside information, the firm may face a lemons market in issuing new
equity.' However, they show that, if a firm can issue default-free debt, then the
firm does not have to pay a premium to outside investors. One conclusion of our
paper, as discussed below, is that firms would be motivated to issue default-free
debt even if there were no information assymmetries at the new issue date.
By focusing on information asymmetries within a trading context, we can
develop a notion of a security's "liquidity." A liquid security has the characteristic
that it can be traded by uninformed agents, without loss to insiders. We show
how intermediation can create liquidity by splitting the cash flows of the under-
lying assets that they hold. By issuing debt and equity securities against their
risky portfolios, intermediaries can attract informed agents to hold equity and
uninformed agents to hold debt which they then use for trading purposes. The
idea that intermediaries can alleviate the problem of trading against insiders
provides a foundation for the demand for a medium of exchange such as money,
which is often simply assumed in many monetary models (e.g., a cash-in-advance
constraint).
Thus, we provide an argument for the existence of intermediation which is
distinct from the previous literature. Recent research on the existence of inter-
mediaries can be broadly divided into two literatures. One literature focuses on
efficient lending arrangements when there exist information asymmetries be-
tween borrowers and lenders. Intermediaries are seen as the unique solution to
such agency problems. Examples of research in this area include Diamond (1984)
and Campbell and Kracaw (1980). Unlike this literature, which focuses solely on
the asset side of intermediaries, our paper is similar to a second line of research
which has investigated the properties of intermediaries' liabilities. In the seminal
paper by Diamond and Dybvig (1983), banks provide liquidity by acting as risk-
sharing arrangements to insure against depositors' random consumption needs.

1 Rock (1986) considers a similar problem.

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 Financial Intermediaries and Liquidity Creation 51

The intermediary contract prevents inefficient interruptions of production.

Like Diamond and Dybvig (1983), we are concerned with the idea that inter-
mediaries provide liquidity. However, our notion of intermediaries as providers
of liquidity differs in a number of important respects. As Jacklin (1987) and Cone
(1983) have shown, a crucial assumption of Diamond and Dybvig (1983) is that
agents cannot trade equity claims on physical assets. If a stock or equity market
is open, this trading arrangement weakly dominates intermediation. Unlike
Diamond and Dybvig, we do not arbitrarily rule out trading in a stock market.
On the contrary, it is the presence of insiders in this market which motivates the
formation of an intermediary. Second, our model differs in that the intermediaries
here will explicitly issue debt and equity, serving as mechanisms that split cash
flows. Finally, the existence of our intermediary does not rely on providing risk
sharing or resolving inefficient interruptions of production. Our notion of liquid-
ity as providing protection from insiders is fundamentally different.
Recent independent work by Jacklin (1988) is similar to ours in that, in the
context of a Diamond and Dybvig-like model, he does not rule out trading in an
equity market and shows that bank liabilities can prevent losses to informed
insiders. However, the intermediary modeled by Jacklin does not issue debt and
equity and is partly motivated on risk-sharing grounds. Our model differs in that
intermediaries explicitly issue both debt and equity securities, thereby splitting
the cash flows of their asset portfolio. Thus, in our setup, intermediaries explicitly
create a new, liquid security. We also consider the feasibility of this intermediary
contract by considering the conditions under which the intermediary can attract
insiders to become equity holders. Thus, we justify the bank from first principles
on grounds different from risk sharing.
Importantly, bank intermediation is not the unique solution for protecting
uninformed agents. In our model, liquidity creation may be accomplished at the
firm level without the need for bank intermediation. By issuing both equity and
debt, firms can split the cash flows of their asset portfolios, thereby creating a
security (corporate debt) which is safer than their underlying assets. This debt
can serve as the basis of a safe security that may be used by uninformed agents
for transaction purposes.
A key point is that private transactions contracts may not be feasible under
certain conditions. This might be viewed as a "market failure" from the perspec-
tive of the uninformed agents and could justify a role for government intervention.
The government can intervene on their behalf in several ways. One way of
protecting the uninformed agents is by insuring the deposits of the banking
system through a tax-subsidy scheme. A system of government deposit insuranc
can achieve the same allocation as when private bank transactions contracts are
feasible. Alternatively, if it is infeasible for corporations to issue sufficient
amounts of riskless debt, government intervention in the form of a Treasury bill
market can improve uninformed agents' welfare by providing additional riskless
securities. This form of intervention is shown to parallel that of the provision of
deposit insurance since, in both cases, the government's role is to create a risk-
free asset.
The paper proceeds as follows. In Section I the model economy is detailed. In
Section II a stock market allocation when all agents are fully informed is set out

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
52 The Journal of Finance

as a reference point. Section III considers the case of asymmetric information

and shows how the informed agents can take advantage of uninformed agents by
forming a coalition that trades in the stock market. Then, in Section IV the
private intermediary contact, when feasible, is shown to break the informed
agents' coalition. When private contracts are infeasible, we show in Section V
that government intervention by insuring bank deposits or creating a government
debt market can be beneficial in protecting uninformed agents. Section VI
concludes.

I. The Model Economy

There are three dates in the model economy, t = 0, 1, 2, and a single consumption
good. The following assumptions detail the model.

Al. Preferences

There are three types of agents:

(i) Agents with known preferences at t = 0, who derive utility from consump-
tion at date t = 2 given by U = C2.
(ii) Agents with preferences that are unknown at date t = 0, but which are
realized at date t = 1 to have utility from consumption at date t = 1 given
by U = C1 but no utility from consumption at t = 2. These agents are
called "early" consumers.
(iii) Agents with preferences that are unknown at date t = 0, but which are
realized at date t = 1 to have utility from consumption at date t = 2 given
by U = C2 but no utility from consumption at date 1. These agents are
called "late" consumers.

Agents of types (ii) and (iii) will collectively be called "liquidity traders." Let N
equal the number of liquidity traders, which is assumed to be large relative to
the number of agents with known preferences. At t = 1 the proportion of liquidity
traders with preferences for early consumption is realized. (The remaining
fraction consists of late consumers.) The realized proportion of early consumers
may be low, proportion wl, which is expected to occur with prior probability ql,
or high, proportion Wh, expected to occur with prior probability qh. It is assumed
that Wh > Wl.

A2. Endowments and Technology

At t = 0, all agents receive endowments of a capital good which when invested

earn a return in the form of the consumption good at t = 2. Each liquidity trader
is assumed to receive an endowment of one unit of the capital good, while type
(i) agents with known preferences receive equal endowment shares of the capital
good that total M units in aggregate. Capital is homogeneous, and each unit
produces the same random return. Each capital unit produces either RH units of
the consumption good or RL units of the consumption good at date t = 2, where
RH> RL > 0. It is assumed that the probabilities at date t = 0 of each state
occurring equal one half.

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 Finlancial Intermediaries and Liquidity Creation 53

In addition to the capital good, all liquidity traders receive an endowment of

e1 units of the consumption good at t = 1, while type (i) consumers receive equal
endowment shares of the consumption good at time t = 2 that total Me2 units in
aggregate. Each unit of the consumption good received by the liquidity traders
at t = 1 can either be consumed at t = 1 or stored to yield a certain return of one
unit of the consumption good at date t = 2.

A3. Information Sets

At date t = 1, uncertainty about capital returns and liquidity traders' prefer-

ences is resolved. It is assumed that type (i) consumers have access to this
information at date t = 1; i.e., they know whether the return on capital will be
high or low and whether the proportion of early consumers in the economy is
high or low. Thus, we will hereafter refer to the type (i) consumers as the
"informed" traders.
While liquidity traders find out at t = 1 whether they are early or late consuming
individuals we will consider the case where they are not directly informed about
the aggregate proportion of early consumers and the realized return on capital.
In this case, information may or may not be revealed by the result of traders'
actions at time t 1. However, for purposes of comparison, we will first consider
the "full-information" benchmark case where liquidity traders are assumed to
directly receive information regarding the realized aggregate proportion of early
consumers and the realized return on capital.

II. A Stock Market with Full Information

It is apparent that certain agents will desire to trade at t = 1. In particular, when

some liquidity traders find that they are early consumers at t = 1, they will want
to sell their entire endowment of the capital good for the consumption good at
this time. In addition, other liquidity traders who discover that they are late
consumers may want to sell their t = 1 endowment of the consumption good for
the capital good if their expected return to holding capital is at least as good as
their return to storing their consumption endowment.In general, the type (i)
informed traders may desire to sell some of their capital good for the consumptio
good at time t = 1 in order to store it from t = 1 to t = 2. Whether informed
traders want to sell capital will be an important issue when we consider the case
of uninformed liquidity traders. However, it will become clear that ignoring the
type (i) traders will riot change the equilibrium for the full-information case.
Since each unit of capital invested at t = 0 is subject to the same source of risk
(i.e., either all units produce a high return or all units produce a low return at t
= 2), it will make no difference whether we think of agents individually investing
their endowment of the capital good or giving it to firms who then issue to them
shares reflecting a proportional claim to the capital's return at t = 2. Thus, a
"stock market" is equivalent to individual investment of the capital good.
Let us then consider the stock market equilibrium in this full-information
case. All agents' utility levels will be determined once we solve for the equilibriu
price of the capital good in terms of the consumption good at date t = 1. We do

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 54 The Journal of Finance

this for the four possible states of natu

H,L, where i refers to a high or low prop
to a high or low return on the capital go
one unit of the capital good in terms of units of the consumption good when state
i,J occurs.
At t = 1 early consumers will wish to purchase the consumption good in
exchange for their endowment of one unit of the capital good. Early consumers,
in total, own Nwi units of the capital good which they are willing to sell. The
aggregate quantity of the endowment good demanded by the early consumers is
Nwipij. Since the late consumers are the only agents from whom the early
consumers can buy endowment of the consumption good, the late consumers will
end up selling some or all of their endowment of the consumption good to the
early consumers. Let the amount of consumption good supplied by the late
consumers be S(pij). If everything is supplied, then S(pij) = Ne1(1 - we).
Otherwise, some amount less than Nel(l - wi) will be supplied.
We now determine the price, Pi>, which clears the market at date t = 1 in each
state of the world I i,j}. Market clearing equates the demand for the consumption
good with supply. Thus,

Nwipij s Ne1(1 - we). (1)

There are two separate cases to consider, one where late consumers sell all of
their consumption endowment (condition (1) holds with equality) and one where
they sell only part, choosing to store some (condition(1) being strict inequality).
When there is no storage in equilibrium, condition (1) becomes an equality.
Solving for the price of the capital good, we have
el(1 - wi)(2
(No Storage) Puj = W * (2)
This case holds under the parametric restriction:

Rj e1(l - wi) (3)

Wi

When storage occurs in equilibrium, late consumers must be just indifferent

between buying and holding the capital good and storing the consumption good,
i.e.,
(Some Storage) Pij = Rj. (4)
This case holds when the inequality sign in condition (3) is reversed.
Hereafter, we will make the assumption that condition (3) holds for j = H, so
that, in equilibrium, no storage will occur for the states Ih, HI and 11, H}, where
the return on capital is high. In addition, we will assume that condition (3) does
not hold for j = L, so that, in equilibrium, some storage will occur for the states
Ih, L and I1, L }, where the return on capital is low. These assumptions can be
summarized by the following condition:

el(1 - w1) el(l - Wh) (5)

RH Wh- > >RL(
WI Wh

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 Financial Intermediaries and Liquidity Creation 55

Condition (5) amounts to assuming a sufficiently high variance in asset returns

relative to the variance in the proportion of early consumers. This assumption
will lead to a more interesting problem when we consider the effects of asymmetric
information.
Note that, for this full information case, type (i) consumers have no incentive
to trade in the capital good at date t = 1. Whenever there is a high return on
capital, the rate of return on capital exceeds that of storing endowment, so type
(i) consumers will choose not to sell capital. When there is a low return on
capital, the rate of return on capital just equals the return to storage, so that
type (i) traders are indifferent to purchasing endowment.
Since type (i) agents do not trade, their expected utility (consumption) per
unit of capital endowment at date t = 0 is

E[C2] = e2 + R, (6)
where R-1/2(RH + RL)

The expected utility of liquidity traders can be computed from our previous
results:

E[C1 + C2J = - IWh(el + PhH) + (1- Wh) RH + P1RH

+ q'-[wI(e1
2
+ PIL) + (1 - wl) (RL + el)]

+ 2h[Wh(el
2
+ PhL) + (1 - Wh) (RL + el)] (7)

[W2
+ ql wi(e1 + PIH) + (1-e
w1) RH+PIH
eR H )RH ,j

= el +R.

In what follows, we will compare the expected utility of the

under alternative information and trading settings to the expected utilities given
by (6) and (7).

III. A Stock Market with Asymmetric Information

Now suppose the model is the same as that of the previous section except that
only type (i) agents, the "informed traders," are assumed to have direct knowledge
of the return on capital and the proportion of early consumers at t = 1. In this
section we restrict liquidity traders to hold their wealth only in the form of stock.
Given this assumption we ask whether the informed agents can collude at date t
= 1 to exploit the liquidity traders. First, we summarize what will happen at t =
1. Then we define an equilibrium. Finally, we show the existence of insider
trading in equilibrium.

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
56 The Journal of Finance

The liquidity traders, early and late consumers, do not know what return
capital goods will earn. Nor do they know the proportion of early consumers in
the economy. At date t = 1, however, the decision of the early consumers is
straightforward. Regardless of possible information, they sell their capital goods
for consumption goods. Late consumers must decide either to store their newly
arrived endowments of the consumption good or to sell all or parts of these
endowments for capital goods. This decision, made as a function of the market
price, characterizes the behavior of the late consumers.
Informed agents know (as do all agents) that, in equilibrium, prices will reveal
some or all information about the true state of the world. Consequently, they
will need to coordinate their trading strategies (collude) in order to gain from
their superior information. We assume that there is a sufficiently small number
of informed agents such that they are able to form a trading coalition, if they
individually so desire. Thus, at t = 1 the sequence of events is as follows. First,
the informed agents communicate and choose an amount of capital goods that
they will jointly supply in state {i, j} knowing that uninformed agents will act
competitively. We first solve this game between the informed agents. Then the
equilibrium price is determined to clear the market between late consumers
supplying endowment goods and early consumers, possibly together with in-
formed agents, selling capital goods.
The amount supplied by the coalition in each state {i, j} will be based on a
strategy designed to make some states of nature indistinguishable from other
states of nature when viewed by the uninformed agents. That is, the equilibrium
prices in some states of nature will be the same as in other states of nature. In
order for prices not to reveal the true states of nature in equilibrium, the optimal
strategies of individual informed agents must be to supply no more capital goods
than are supplied by the coalition acting on their collective behalf. The existence
of the insider trading equilibrium will depend on showing that individual members
of the informed agents' coalition have no incentive to deviate from the coalition
strategy, by selling capital goods on their own unbeknownst to the coalition. In
equilibrium it will be in the interest of each informed agent to be a member of
the coalition and, once having committed capital for sale by the coalition, not to
supply any additional capital. This is because, if any additional capital is supplied
by individual informed agents (acting independently of the coalition), the equi-
librium price will reveal the true state of the world. If this occurs, then no
informed agent can benefit. We now briefly formalize this so that we can
subsequently define an equilibrium.
Let Mij ' M be the amount the coalition proposes to its members as the
amount to be supplied in state { i, j}, with each member supplying an identical
share. The coalition's strategy will be characterized by the amount of the capital
good that the coalition supplies in state { i, j}, Mij. We say that Mij is a self
enforcing Nash coalition in state {i, j} if any subcoalition of informed traders,
taking the capital supplied by the complement of the subcoalition as given,
chooses to abide by the per capita shares assigned by the whole coalition. If this
is true for all possible subcoalitions, then the coalition Mij is not subject to
collapse since there is no incentive for any member or group of members to

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 Financial Intermediaries and Liquidity Creation 57

deviate from the proposed mij.' We will refer to this coalit

Coalition."
Market clearing will require that the price, say p*, equate the demand for
consumption goods with the supply of consumption goods in state {i, j}:

Nwip7* + Mijp, = S(p*). (8)

As before, the supply, S(pi0), will be either all the endowments of the late
consumers, N(1 - wi)e1, or some lesser amount if there is storage in equilibrium.
We now define a Nash-type equilibrium in this setting. An Imperfectly Com-
petitive Rational Expectations Equilibrium is (a) a price system, {pij}, (b) speci
cation of storage strategies for the late consumers, S(p0j), and (c) a specific
of insider coalition strategies, {Mij}, such that, given {pij>, knowledge of the
model, and the information set of the informed agents in state { i, j}I, the storage
and coalition strategies of the respective agent types are chosen such that (i)
their respective utilities are maximized, (ii) IpijI clears the market in sta
and (iii) I Mij} is self-enforcing.
Let R* = qh'Rh + ql'RL be the uninformed late consumers' expectation at time
1 of the return on capital when state 11, LI actually occurs, where qh' and q
their posterior probabilities of the states being wi = Wh and wi = wi, respect
The following proposition demonstrates the existence of insider trading by the
informed agents.

PROPOSITION 1 (Insider Trading): Let {Pi>} be the full-information prices for states,

{i,j}. If (i) e1(1 - wh) /wh c R* and (ii)-> (W l W_ then there exists Imperfectly
N (- Wh) '
Competitive Rational Expectations Equilibrium prices {pj }, where PIH = H, PhL
PhL, and PhH = PIL = pAhH. That is, these prices are fully revealing in only two of the
four states.

Proof: We will verify that the following specification of prices and strategies
constitutes an equilibrium for the assumed parameter values.

State {1, HI

PIH - ; MIH = 0; S(pI) = N(1 - wj)e1 (No Storage).

Wl

State { h, LI
PhL = RL; MhL = 0; S(PhL) < N(1 - wj)ej (Some Storage).

State {h, HI
*_el(l - Wh)
PhH - ; MhH = 0; S(PH) = N(1 - wj)ej (No Storage).
Wh

2 See Bernheim, Peleg, and Whinston (1987) for the motivation for this definition of a se
enforcing coalition. This equilibrium concept refines the set of possible Nash equilibria of the game
between the insiders when they choose the Insider Coalition strategy. For our purposes it focuses
attention on equilibria of interest, namely ones in which insider trading occurs.

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 58 The Journal of Finance

State I 1, LI

* = el(l - Wh) N(Wh - WO )(

PiL MLe (1 - Wh)S(P
Wh(1- ) = N(1 - wh)el (No Storage).

The proposed equilibrium pric

HI, are the full-information pr
revealing and are market clearing. It remains, then, to show that the actions of
the insider coalition can cause prices to only partially reveal information in the
states Ih, H} and I1, L}.
In state I1, L }, the return on the capital goods is low, and informed agents
would like to sell their capital goods in exchange for consumption goods at the
assumed equilibrium price. They will then store the consumption goods for one
period. Since the proportion of the late consumers is low, wl, the informed
coalition can mimic the state Ih, HI where there are many late consumers and
the informed agents don't enter the market.
Thus, if the late consumers supply all their endowment of consumption goods,
then market clearing requires

NwIP* + MiLP* = N el (1 - wl). (9)

(1- Wh)
Now, set P*H = P*H = el and solve for MIL:
Wh

N(Wh - Wl)
ML=(1 - Wh) (0

Condition (ii) of the proposition insures that insiders have sufficient capital for
(10) to hold. By supplying MIL units of the capital good in exchange for the
endowment good, the insider coalition can create the false impression that the
state is {h, H when, in fact, the state is I1, L }, However, for this to be successful,
two further considerations need to be examined.
First, will late consumers choose to sell their endowment when they see the
market clearing price p* ? They will if, on average, it is profitable to do so, i.e
when condition (i) of the proposition holds:

PIL= e1Wh c <R* = qh'RH + ql RL. (11)

If 1,te consumers form their expectation of the state being 1, LI or Ih, HI in a

Bayesian fashion, conditional on the fact that they, themselves, are late con-
sumers, then

(q1 - Wh)
qh =h ~qh(l - Wh) + ql(l - WX

ql' = q, (1 - wi)
qh(l - Wh) + ql(l - w1)

Condition (11) says that, even though late consumers know that the informed
coalition will cheat them in state 1, LI and that this cannot be detected, still it

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 Financial Intermediaries and Liquidity Creation 59

is optimal to sell all their endowment. It is optimal if qh' is sufficiently large, so

that most often the true (but unobserved) state is Ih, H}.
Second, we must check that MIL is a self-enforcing Nash coalition. If there is
a total of M units of capital owned by the informed agents, and they are all in
the coalition, then each can exchange MiLIM per unit of the capital for endow-
ment goods. Note that, if any member or group of members independently
demands additional endowments, then the market clearing condition (9) will not
hold at pI and the new price will reveal the collusion. Uninformed agents will
infer the truth. If the state 1, LI is revealed, late consumers will not be willin
to sell their endowments. If there is a deviation from MIL, then the informed
agents as a group will not benefit, including the member or group who deviated.
Therefore, since any deviation results in a fully revealing price and, hence, no
benefits to informed agents, MIL is self-enforcing. Q.E.D.
We can now calculate the expected utility per unit of capital endowment for
the informed traders. While M is the total amount of capital endowment of the
informed agents, the coalition can only sell MIL units in state U1, LI. Therefore

E(C2) = e2 + H + 2+ q2 [RL + Wn(PlRL)I -1

2 2 2 (12)

= e2 + R + -Wm(PIL RL),
2 1

where wm - MIL N(Wh - Wl)

M M(l-Wh)

Since RL < pI, by assumption (5), the expected utility of an informed trader
exceeds the full-information expected utility since Wm > 0.
Likewise, we can calculate the expected utility of liquidity traders. It is
straightforward to show that

E[C1 + C2] = el + R - q (Wh -W) [PI -RL] (13)

2 (1 - Wh) 1
Note that this utility is less than that of the full-information case. We now
turn to investigating whether the liquidity traders can prevent being victimized
by the informed traders.

IV. Private Liquidity Creation

In the previous section, liquidity traders were not allowed to contract. The result
was the existence of insider trading that increased the welfare of informed traders
at the expense of the liquidity traders. We now allow the liquidity traders to
respond by contracting. We show that allowing liquidity traders to contract can
prevent insider trading by breaking the informed agents' coalition; i.e., the insider
trading equilibrium analyzed in the previous section will no longer exist. Next,
we show that an alternative equilibrium characterized by bank intermediation
can exist. Finally, we show that the allocation achieved with the bank can be
replicated at the firm level with corporations issuing riskless debt.

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 60 The Journal of Finance

A. Bank Intermediation and Liquidity Creation

Suppose at date t = 0 the following contract is offered to agents. An interme-

diary will be set up which pools agents' capital and issues securities to them. Let
A = N, + MI be the total endowment of the capital good contributed by member
of this intermediary as of date t = 0, where N1 = N - Ns and MI = M - Ms. The
subscript I refers to the capital of agents joining the intermediary, and S refers
to the capital of agents continuing to invest in the stock market. The total return
of the intermediary's assets at date t = 2 is ARi, i = H, L. Ownership of two types
of claims on this capital is offered to agents: debt claims and equity claims. Let
D and E (whose sum equals A) be the total amount of capital contributed by
agents who own debt and equity claims, respectively.
The contract also imposes a debt-to-equity ratio ceiling such that the total
payment promised to debt claim, DRD, must be less than or equal to ARL; i e.,
debt claims are required to be riskless:

DRD 'ARL = (D + E)RL. (14)

Therefore,

(D < E RL or E D(RD- RL) (15)

(D +EJ-RD R
We would like to consider whether offering agents this intermediary contract
would affect the Imperfectly Competitive Rational Expectations Equilibrium
analyzed in the previous section. Before stating a series of propositions related
to this issue, we make an additional assumption that will simplify the proof of
the first of these propositions. We assume that conditional on being a late
consumer, the probabilities of the state being wi = Wh or wi = w, are equally
likely. If late consumers form expectations in a Bayesian manner, this implies

qh(l - Wh) = ql(l - w1). (16)

Now suppose that liquidity traders are allowed to offer the intermediary
contract to all agents as a possible trading mechanism. It is clear that, for RD
sufficiently high, liquidity traders are better off holding bank debt. The question
is whether the informed agents can be induced to defect from the Insider Coalition
to become the bank's equity holders. If this occurs, the intermediary contract
will be feasible and the equilibrium of the previous section will not exist.

PROPOSITION 2 (Nonexistence of Stock Market Insider Equilibrium): Consider a

small number of liquidity traders, say N, (close to zero), choosing to form a bank

Then, if the ratio of informed to uninformed agents'capital, N is sufficiently large,

there exists a rate of return on intermediary debt, RD, such that (i) debt is riskless,
(ii) liquidity traders prefer to invest their capital in the debt of the intermediary
rather than the stock market, and (iii) individual informed agents prefer to invest
their capital in the equity of the intermediary rather than the stock market insider
coalition.

Proof: See the Appendix.

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 Financial Intermediaries and Liquidity Creation 61

Proposition 2 provides a condition under which individual liquidity traders and

informed agents have an incentive to form an intermediary at time 0 rather than
invest in the stock market. The higher the ratio of total capital of the informed
agents to that of the liquidity traders, the smaller will be the expected profits of
the informed agents in the Insider Coalition. Therefore, the higher this capital
ratio, the smaller is the required rate of return on bank equity necessary to induce
an individual insider to join the bank and defect from the Insider Coalition.
Consequently, if the required return on bank equity is not too large, the rate of
return on bank debt will be large enough to attract an individual liquidity trader
away from the stock market as well.
The next proposition states that an equilibrium can exist where all liquidity
traders choose to purchase the riskless debt of an intermediary and informed
agents derive no advantage from operating an Insider Coalition in the stock
market. The proof of this proposition assumes the following condition, which
includes condition (5) assumed previously:

R e(l - Wh) - e1l (1-wi)

RL <e( Wh <<RH. (17)
Wh Wl

PROPOSITION 3 (Exis
N
large, then there exists an equilibrium where (i) all liquidity traders purchase
riskless debt of the intermediary and (ii) informed agents will choose to contribute
equity capital.

Proof: See the Appendix.

The intuition behind this result is that, if informed agents' capital is sufficiently
large relative to that of the liquidity traders, it is feasible for a bank to issue
sufficient riskless debt that can be used by all liquidity traders for transactions.3
Implicitly, the existence of this bank contract allows the informed agents to be
identified so that trade with them can be avoided. All liquidity traders who are
early consumers will trade bank debt for endowment at date t = 1. Late consumers
considering selling their endowment at date t = 1 will never choose to purchase
stock market capital because they know that only informed agents will be
supplying stock market capital for endowment, and then only when the return
on capital is low. Thus, the stock market becomes an Akerloff (1970) "Lemons"
market, and late consumers will choose to trade only with early consumers selling
intermediary debt. In this sense, liquidity traders are able to "protect" themselves
from possible disadvantageous trades with the better informed agents.
In summary, we have shown that conditions exist where liquidity traders are
better off holding intermediary debt which is made riskless because some in-
formed investors will voluntarily contribute equity capital for the intermediary.
Under these conditions, with NI = N and NS = 0, the advantage that the Insider

'In addition, as is shown in the Appendix, the greater RL is, the higher is the feasible leverage of
the intermediary, i.e., the smaller is the proportion of informed agents needed to join the intermediary
to make its debt riskless. The greater the leverage, the less RD needs to be lowered in order to raise
the expected rate of return on the intermediary's equity in order to attract informed agents.

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
62 The Journal of Finance

Coalition derives from superior information is completely eliminated. With no

one to trade with at date t = 1 except other informed agents, informed agents'
expected rate of return on stock is simply R. With sufficient defections from the
Insider Coalition, the competitive expected rate of return on intermediary equity
will also approach R, resulting in a deposit rate, RD, with a limiting value equal
to R. Hence, the private intermediary contract can result in an allocation which
gives all agents an expected utility arbitrarily close to the full-information case.

B. Corporate Debt and Liquidity Creation

So far we have implicitly assumed that "firms" do not issue debt. That is, when
we considered the stock market equilibrium in Section III, we imagined individ-
uals exchanging their capital with firms who issued them equity shares. In this
section we briefly consider what happens if the firms are willing to buy capital
at t = 0 in exchange for either debt or equity. So now there exists a market for
corporate debt, such as commercial paper.
Suppose a firm offers to pay RD per dollar of debt and issues an amount of
riskless debt such that DRD = ARL, where A = D + E is the firm's total assets.
Then it is immediately apparent that the firm can offer the same riskless debt
as the bank intermediary we described previously. All of the above arguments
about the bank now apply to the firm. Agents need not directly hold the claims
of firms, but mutual funds could arise to specialize in holding either debt or
equity claims. In particular, funds similar to money market mutual funds could
purchase the high-grade debt (e.g., commercial paper) of firms. As before, the
equilibrium would be for all liquidity traders to buy claims on the debt fund and
all informed traders to buy claims on the firm's equity. We comment further on
this in our concluding remarks.

V. Deposit Insurance and a Government Debt Market

A deposit insurance system for banks can also satisfy the liquidity traders' desire
for a safe asset for trading. In this section we show how deposit insurance can
replicate the allocation of the previous section when intermediary debt is risky.
In addition, we show that development of a government debt market is similar
to deposit insurance, as it involves government creation of a risk-free asset. In a
like manner, a government debt market can replicate the riskless corporate debt
contract when riskless corporate debt is in insufficient supply. The government
can succeed where private contracting fails due to its ability to enforce lump sum
taxation. It is the revenue from this taxation that accounts for the government's
ability to create riskless securities.
As Merton (1977) has observed, "the traditional advantages to depositors of
using a bank rather than making direct market purchases of fixed-income
securities . .. economies of scale, smaller transactions costs, liquidity, and con-
venience ... are only important advantages if deposits can be treated as riskless."
Presumably, if deposits were not riskless, then small agents would face infor-
mation and surveillance costs necessary to evaluate the current risk of bank
liabilities. Without this information, other informed agents might then take

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 Financial Intermediaries and Liquidity Creation 63

advantage of them. Consequently, less informed agents would benefit if there

were deposit insurance. Indeed, a stated goal of government deposit insurance is
to protect the small investor.
Suppose that deposits are risky, i.e., DRD> ARL. This would be the case if, for
example, the capital endowment of the informed agents were too small to provide
enough riskless debt or if RL = 0. In other words, if the low return rate state of
the world is realized, then deposits will incur a capital loss. The insurance system
works as follows. If RL is realized, so that the bank would not be able to meet its
promised payments at time t = 2, then the government is assumed to tax all late
consuming agents in proportion to their endowment in order to raise enough
revenue to pay off the bank debt at par.4 The government will also charge an
insurance premium at time t = 2 that the bank pays if it does not fail, i.e., when
RH is realized, which is allocated to all late consuming agents.
Let T be the tax revenue collected when the bank fails. In order to avoid a
capital loss on deposits if RL is realized, the amount of insurance needed is T =
DRD- ARL. Each agent consuming at date t = 2 pays a share of T. At t = 2 there
are informed agents who were endowed with M units of capital and N(1 -wi),
wi = w, or Wh, late consuming liquidity traders, each having been endowed with
one unit of capital. This insurance arrangement will only be feasible if, regardless
of the proportion of early consumers, the remaining agents can afford to pay the
tax. Thus, feasibility requires

T/[M + N(1 - w)] < e2, i = 1, h, (18a)

RD
T/[M + N(l1- wj] < RD + el D, i = 1, h. (18b)
PDi

Informed agents have, at least, Me2, their second-period endowment.5 Thus, the
tax per unit capital cannot exceed the e2 endowment. This is requirement (18a)
above. Similarly, (18b) requires that the late consuming liquidity traders, who

have assets of RD + e1 D, be able to afford the tax. (The values of PDi are gi
PDi
by (A18) in the Appendix.)
If the bank does not fail, then the bank pays an insurance premium of X to the
rest of the economy, which consists of all informed agents and depositors. The
expected return to the bank equity holders in the presence of deposit insurance
is

E[RE]E = (1/2)[RH(D + E) - (RD + 4)D] + (1/2)0. (19)

It is straightforward to solve for a fair insurance premium. Since bank failure

and bank solvency are equally likely, i.e., RL and RH each occur with probability
one half, a fair insurance premium equates the amount paid as a premium in the
high state with the amount of insurance coverage in the low state:

OD = T= DRD-(D + E)RL, (20)

'The government is assumed to observe the bank failure at date t = 2.

5 Informed agents holding bank equity have only e2 per unit of initial endowed capital since their
bank equity is worthless, while informed agents in the stock market have e2 + RL.

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
64 The Journal of Finance

which implies that

0 RD _(D + E)R (21)

Substituting (21), the expression for the fair deposit insurance premium, into
(19) yields

E[RE]E = R(D + E) - RDD. (22)

As in the previous section, consider a competitive equilibrium where the expected
rate of return on equity approaches R. In this case, equation (22) shows that RD
also approaches R. Therefore, the allocation under the deposit insurance scheme
gives agents the same expected utility as in the case of the private uninsured
intermediary considered in the previous section. In summary, we have shown the
following.

PROPOSITION 4 (Deposit Insurance): When bank debt is risky, the tax-subsidy

scheme IT, (}, defined above, can implement an allocation which gives all
the same expected utility as in the riskless private bank deposit allocation.

Similar to government intervention as a deposit insurer, we can consider

whether government intervention can benefit uninformed agents when firms
issue corporate debt, as was described previously. Let us start from the assump-
tion that each firm issues riskless debt such that

A,RL ' DiRD, (23)

where Ai and Di are the assets and debt of firm i, respectively. However, suppose
that the assets of firms are of sufficient risk to preclude uninformed agents from
placing their entire wealth in risk-free corporate debt. In this case, government
intervention in the form of a government debt market can allow uninformed
agents to replicate the allocation of the previous Section IV.B, where riskless
corporate debt was in sufficient supply.
As with the deposit insurance scheme, the government can create additional
two-period risk-free securities backed by lump sum taxation of agents' endow-
ment in period 2. The government simply issues claims on second-period endow-
ment equal to the difference between uniformed agents' time 0 endowment and
the supply of risk-free corporate debt, so that the government sells bonds for
capital equal to N - D at time 0. Since government and firm debt are perfect
substitutes, they each pay a two-period return of RD, implying that the time t =
2 maturity value of government bonds B equals

B = (N - D)RD. (24)

The government is assumed to invest the capital it acquires at time 0, either

directly investing it itself or giving it to firms which issue it equity shares. At
time t = 2, this investment is worth (N - D)Ri, i = H, L. The short fall (excess)
between this investment return and the promised payments on bonds, B, is made
up by lump sum taxation (subsidization) of late consumers, subject to feasibility
requirements similar to (18a) and (18b). Competitive equilibrium implies that
the expected return on equity as well as the return on riskless debt will equal R.

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 Financial Intermediariies and Liquidity Creation 65

Thus, the additional debt supplied by the government can allow uninformed
agents to purchase sufficient risk-free securities to meet their demands for
liquidity. Hence, this intervention can also restore the uninformed agents an
allocation which gives them the same expected utility as in the full-information
case.

VI. Conclusion

The historically popular notion that informed agents can benefit at the expense
of uninformed agents is true in the setting which we have analyzed. Informed
agents can form an insider coalition which is self-enforcing and can benefit at
the expense of the lesser informed agents. When this condition exists, a demand
for liquid securities by uninformed agents will result. By splitting risky cash
flows, liquid securities are created which have the effect of eliminating the
potential advantage possessed by better informed agents.
Liquidity can be created through the formation of banks. We have formalized
a traditional rationale for the existence of banks and deposit insurance, namely
that they provide a riskless transactions medium that eliminates the need of
uninformed agents to trade in assets whose returns are known by better informed
agents. By issuing deposits, banks create "riskless" securities for trading purposes.
In instances where bank asset risk is such that uninsured deposits cannot be
made riskless, we have shown that deposit insurance can replicate the allocation
achieved with riskless private bank deposits.6 In addition, liquid securities can
also be created through the formation of corporate debt or government securities
markets. As an alternative to bank intermediation, firms can split risky cash
flows, thereby creating a safer security (debt).
An empirical implication of our model is that transactions securities should be
the most actively traded assets. This is consistent with the relatively high
turnover in ownership of insured bank liabilities and Treasury securities. Cor-
porate debt, on the other hand, is much less actively traded, suggesting that our
assumption that firms can create riskless securities simply by splitting the cash
flows of their underlying assets is not completely accurate.
For tractability, we studied a model with a single source of asset risk. Clearly,
with multiple sources of asset risk, diversification would provide another, perhaps
complementary, channel for the reduction of risk. This channel implies combin-
ing imperfectly correlated assets to reduce risk, rather than splitting cash flows.
These issues are investigated in Gorton and Pennacchi (1989). The creation of
mutual funds holding a diversified portfolio of corporate debt can alleviate the
inability of individual firms to create riskless debt. For example, money market
mutual funds are large holders of commercial paper, and the shares of these
funds provide a potentially important transactions medium.7

6An issue which we have not considered concerns possible equilibria where banks exist but their
uninsured bank deposits are risky. In this situation we conjecture that the liquidity traders would be
better off than without the bank but clearly would not be as well off as the case of riskless bank debt.
The value of risky bank debt would depend on the state of nature, but to a lesser extent than would
stock. Informed traders might still use their information advantage.
7Currently, the transactions services provided by money market mutual fund shares may by
inhibited by regulation which denies these mutual funds independent access to the payments system.
Money market mutual fund check and wire transfers must be carried out through commercial banks.

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 66 The Journal of Finance

Due to the recent growth of the market for short-term corporate debt, the
possibility of substituting money market mutual fund shares for bank debt is
intriguing.8 A public policy debate has smoldered around whether such alternative
instruments should be encouraged or restricted as transactions media. In our
analysis there is not reason to prefer bank debt over money market mutual funds.
However, extending our analysis to consider the regulatory distortions. and
monitoring costs associated with bank deposit insurance might lead to a prefer-
ence for a money market mutual fund-based transactions system.

Appendix

Proof of Proposition 2: Step 1 of the proof is to consider the situation of the

liquidity traders. Given the feasibility of the intermediary, we derive the condi-
tions under which they are better off purchasing the intermediary's debt rather
than investing their capital in the stock market. Step 2 considers the informed
agents and shows that, under the conditions derived in step 1, they may be
individually better off by becoming equity holders in the intermediary rather
than being members of the Insider Coalition that operates in the stock market.
Thus, if informed agents are willing to contribute equity capital, the intermediary
contract is feasible.

Step 1: Let PDij be the number of endowment units received in exchange for
one unit of the debt claim at date t = 1 when the state is { i, j}, where i = 1, h, and
j = L, H. Because of the risk neutrality of uninformed agents, at time t = 1 it
must be the case that

RD Re
= - = rij, (Al)
PDij Pij

where Re is the uninformed late consumers' expectation at time t = 1 of the

return on the capital good at time t = 2 and rij is defined to be this common
expected reinvestment rate when state { i, j} occurs.
We can now calculate the time t = 0 expected utility of an uninformed agent
who invests capital in the stock market, Es[Cj + C2], and the utility of an
uninformed agent who invests capital in the debt of the intermediary, EI[C + C2].

Es[C1 + C2] = , q-(wi(e1 + pij) + (1w-i)rij(el + pi>)), (A2)

{i, j} 2

EA[C1 + C2] = {i,jj

E qi 2
(wi(el + PDii) + (1 wi)rij(el + PDij)). (A3)

The difference between (A3) and (A2) will determine whether uninformed
agents have an incentive to invest in the intermediary.

EI[C1 + C2] - ES[C1 + C2] = E 9'(PDij Pij)(Wi + (1 - w)ri). (A4)

{i, j, 2

Perhaps an unplanned benefit of large government budget deficits has been an increased supply
of riskless debt, further adding to the feasibility of a transactions system backed by money market
instruments.

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 Financial Intermediaries and Liquidity Creation 67

To determine the sign of (A4), we need to compute the prices

in Section III of the text, these prices will, in general, depend on the parameters
of the model as well as the actions of the informed agents. Analogous to condition
(5) in the text, we state the following conditions:

RH > el(l - w1) + N el(1 - w1) - RD (A5)

Wl NS Wl

R el(l Wh
- Wh) + NRel(l DWh) RD. (A6)
NS Wh

Note that, for NI sufficiently small relat

will hold if condition (5) holds. Thus, we wish to examine the incentives for a
small group of uninformed agents to join as intermediary, given that there
currently exists a large number in the stock market.

Analogous to the results of Section III, if conditions (AS) and (A6) hold, then
states Il, H} and Ih, L} are fully revealing, while an Insider Coalition can form
to purchase endowment in state I1, L } to mimic the prices of all securities in state
{h, H}. Using (Al) and equating demands and supply of the endowment good
lead to the following set of state-contingent prices and time t = 1 reinvestment
rates:

e1( - w1)N RD RH RD
(No Storage) PDlH = l-wlNR lH=PI H 1H- D
Wl(NIRD + NSRH) RDP PDIH

(Storage) PDhL = RD, PhL = RL, rhL = 1,

(No Storage)

el(l - Wh)N RD R
PDIL = PDhH +
Wh(NIRD = NsR
hNR NSRA'
) RD (A7) P
RD
rIL = rhH = -
PDIL

where R* is the late consumers' expectation at time

date 2, when the state is only partially revealed to
formula for R* is given in equation (11) of the text
reinvestment rates into (A4) and simplifying, one o

EA[C1 + C2J - ES[C1 + C2J =

- r Wh)N el(qhwh + qlwi)
1/2(RD- R*)[(L Wh(NIRD + NsR ) + qh(l - Wh) + ql(l - wl) (A8)

+ 1?2(RD- RH)ql(l - wl) LNRD+esRH + + 2(RD- RL)qh

It is straightforward to verify that (A8) is a strictly increasing function of RD

and, for RD sufficiently large, uninformed agents will prefer joining the interme-
diary. Furthermore, we can also show that there exists a value of RD < R for

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 68 The Journal of Finance

which (A8) will be positive when all uninformed investors initially invest in the
stock market, i.e., when N1 is small. Taking the limit as NI -* 0 (or NS -* N),
lim EI[C1 + C2]- ES[C1 + C2] =

1/2(RD- R-)L R, e1(qhWh + qlwl) + qh(l - Wh) + q/(1 - w1)j (A9)

+ 1/2(RD - RH) ql(l - w1)(RH)

(eR + 1) + ?/2(RD - RL) qh.

Setting the right-hand side of equation (A9) to zero, we can solve for the
minimum return on intermediary debt, Rm, for which uninformed agents are
well off joining the intermediary as they are staying in the stock market. For the
simplifying case of condition (16), that, conditional on being a late consumer,
the probability of the state being h or 1 is equally likely (R*-R), we have

Rm= R - ql(1 - wl) (R2R)[whej/] (AlO)

where

-Wh) el(q + e qh
0 - R (qhWh+ qlwl) + ql(1 wl) (3 R +qh > O

The term in brackets on the right-hand side of (AlO) is strictly positive be

of condition (5). Since (A9) is continuous and striptly increasing in RD, it
also be strictly positive for some value of RD less than R.

Step 2: Given that liquidity traders have an incentive to leave the stock market
and join the intermediary for RD > R7m, we now show that the intermediary
contract will be feasible if informed agents can be induced to provide equity
financing rather than invest their capital with the stock market Insider Coalition.
The informed agents who are members of the stock market Insider CQalition
will sell their capital to mimic the state {h, H} when the state is actually {1, L}.
They purchase endowment in the amount:

MIL iL
( (-Wh)
- )(NS + NIRp/R*), (All)
which results in their time 0 expected utility per unit capit

EFC2] = e2 + R + q9 M (Pl-A RB, (A12)

where PIL is given by (A7).

Note that, for RD < R*, MIL is less than its value for the case where NI = 0,
which was analyzed in Section III, while PiL is less than pI given in Section III.
Thus, the expected utility of the informed agents falls in this case if MS stays
the same. Now if some informed agents defect from the Insider Coalition and

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 Financial Intermediaries and Liquidity Creation 69

invest their capital, equal to MI, in the equity of the intermediary, their expected
return will be

E[MIRE] = R(NI + MI) - RDNI. (A13)

If the intermediary's capital constraint is binding so that NI and MI follow the

debt and equity proportions given in equation (15), then the expected return on
intermediary equity equals

E[RE] = R + (R - RD)(RD L) (A14)

Thus, comparing (A14) with (A12), we see that an informed agent who invests
in the equity of the intermediary will have a higher expected return than an
informed agent in the Insider Coalition if

(R - RD) qi (NS + NIRD/R*) (Wh - W)

(RD- RL) 2 MS (1 - Wh)

Consider the incentive for informed investors to defect from the stock market
coalition when initially NS is close to N. Taking the limit of (A15) as N, goes to
zero and rearranging terms result in

(R(R
- RD)>2
RD) > M
( -(lW)
wI [WhRL
) [ W - 1] (RD - RL) (A16)

Now suppose RD is set such that R > RD 2 Rm, where RDm is given by (A10).
Then both sides of condition (A16) are strictly positive, but the right-hand side
of (A16) can be made sufficiently small for M sufficiently large. (Note that RDm is
independent of M.) Thus, for MIN sufficiently large, a return on intermediary
debt can be offered which gives both uninformed and informed agents the
incentive to start an intermediary.

Proof of Proposition 3: We first take the feasibility of the intermediary for N

- N as given and later show that this holds for M/N sufficiently large. If all
liquidity traders initially invest in the riskless debt of the intermediary, consider
the possibility of the informed traders being able to strategically purchase the
endowment of the late consumers when the return on stock market capital is
low.
Given condition (17), consider a return on intermediary debt, RD, such that

( Wh) < RD R (A17)

elwh

Similar to the analysis of Section II in the text, it is straightforwar

a full-information equilibrium would result in the time t = 1 prices of i
debt equal to

(Some Storage) PDuj =RD, R = L, H,

(No Storage) PDhj = el(1 - Wh)/Wh, j = L, H. (A18)

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
70 The Journal of Finance

In other words, some storage occurs whenever there is a low proportion, wl, of
early consumers, and no storage occurs whenever there is a high proportion, Wh,
of early consumers. In equilibrium, the price of stock market capital will satisfy

Pij = PDijE[Rj]/RD
iRD = PDRj- (A19)
Now consider the case of asymmetric information. Stock market insiders would
like to be able to purchase endowment and sell stock market capital at time 1
when the return on capital is low, RL. Potentially, they could do this, as before,
when state {1, L} occurs, by purchasing endowment from late consumers. How-
ever, rational late consumers would never choose to sell their endowment for
stock market capital because the only sellers of stock market capital are informed
agents, who the late consumers know would only choose to sell capital when the
return is RL. Unlike the situation considered in Section III, where liquidity
traders invested in the stock market at time 0, late consumers will now realize
that they will only be trading capital with informed agents, and then only when
the return on capital is RL. Hence, late consumers will only offer a price for stock
market capital of

Pij = PDRL. (A20)

iRD.

At this price, there would be no incentive for informed agents to purchase

endowment. Since late consumers would only sell endowment for the riskless
debt of early consumers, PDij would always be equal to its full-information price
given in (A18). This results in the expected utility of uninformed agents being
equal to

E[C1 + C2] = el + RD (A21)

and the stock market Insider Coalition being devoid of power, their return on
capital simply being equal to R. Hence, in order to attract informed agents to
contribute to the intermediary, RD need only be an arbitrarily small amount less
than R, and uninformed agents' utility would approach their full-information
level. In addition, it is straightforward to show that individual liquidity traders
would never choose to invest their capital in the stock market rather than the
intermediary since, if they turn out to be an early consumer, they can only sell
their capital to late consumers at a price which always reflects the return on
capital being RL given by (A20).
Finally, to show that this equilibrium is feasible, informed agents must have
sufficient capital in order to purchase the minimum amount of intermediary
equity required to make the intermediary's debt riskless. Using condition (15),
with D = N we have

M > (RD-RL) (A22)

N RL(A2

Note that the larger RL is, the smaller is the amount of equity capital needed to
enable the intermediary's debt to be riskless.

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms
 Financial Intermediaries and Liquidity Creation 71

REFERENCES

Akerloff, G., 1970, The market for lemons: Qualitative uncertainty and the market mechanism,
Quarterly Journal of Economics 84, 488-500.
Bernheim, B. Douglas, Bezalel Peleg, and Michael Whinston, 1987, Coalition-proof Nash equilibria,
Journal of Economic Theory 42, 1-12.
Campbell, T. and W. Kracaw, 1980, Information production, market signalling and the theory of
financial intermediation, Journal of Finance 35, 863-881.
Cone, Kenneth, 1983, Regulation of Depository Institutions, Ph.D. Thesis, Stanford University.
Diamond, D., 1984, Financial intermediation and delegated monitoring, Review of Economic Studies
51.
and P. Dybvig, 1983, Bank runs, liquidity and deposit-insurance, Journal of Political Economy
91,401-419.
Gorton, Gary and George Pennacchi, 1989, Security baskets and index-linked securities, Working
paper, The Wharton School, University of Pennsylvania.
Grinblatt, Mark S. and Stephen A. Ross, 1985, Market power in a securities market with endogenous
information, Quarterly Journal of Economics, 1143-1167.
Jacklin, Charles, 1987, Demand deposits, trading restrictions, and risk sharing, in Edward D. Prescott
and Neil Wallace, eds.: Contractual Arrangements for Intertemporal Trade (University of Min-
nesota Press, Minneapolis).
,1988, Demand equity and deposit insurance, Mimeo, Stanford University.
Kyle, Albert S., 1985, Continuous auctions and insider trading, Econometrica 53, 1315-1336.
Merton, Robert C., 1977, An analytic derivation of the cost of deposit insurance and loan guarantees,
Journal of Banking and Finance 1, 3-11.
Myers, Stewart and Nicholas Majluf, 1984, Corporate financing and investment decisions when firms
have information that investors do not have, Journal of Financial Economics 13, 187-222.
Rock, Kevin, 1986, Why new issues are underpriced, Journal of Financial Economics 15, 187-212.

This content downloaded from

178.189.189.124 on Sun, 21 Mar 2021 16:54:41 UTC
All use subject to https://about.jstor.org/terms