In this formulation the objects

of choice are not derivative

statistical measures of the
probability distribution of
consumption opportunities but
rather the contingent consump-
State Preference
tion claims themselves set out
in extensive form.
Theory
—J. Hirshleifer, “Efficient Allocation
of Capital in an Uncertain World,”
American Economic Review, May 1964, 80.

F INANCE DEALS WITH INVESTMENT DECISIONS of individuals and firms linked

through the supply and demand for securities in the capital market.1 Firms borrow capital
for investment in real assets by selling securities; individuals obtain claims to firms’ real
assets by investing in securities. Thus securities present opportunities for intertemporal shifts of
consumption through the financing of productive activities. Individual consumption/investment
decisions that determine aggregate security supply are both affected by security prices. By equating
security supply and demand, security prices yield a consistent set of firm and individual investment
decisions. In this chapter, we will analyze how optimal individual investment decisions and optimal
firm investment decisions are determined under uncertainty for a given set of security prices.
Under specified conditions, individual decision making under uncertainty is accomplished
by maximizing expected utility of end-of-period wealth. This decision criterion was shown to
be valid when individuals are rational, prefer more wealth to less, and follow the five axioms
of choice under uncertainty. Implicitly, it was also assumed that individuals can assess a security’s
probability distribution of end-of-period payoffs. It was shown that the expected utility criterion
is a very simple way of choosing among mutually exclusive investments having different probability
distributions of end-of-period payoffs. By choosing the investment with the highest expected
utility, the optimal investment is determined, thus condensing a choice across N probability distributions
of end-of-period payoffs into a comparison among N expected utility values.
In this chapter, we wish to move beyond the individual’s choice problem of mutually exclusive
investments to the more general problem of portfolio decision making, that is, the optimal choice
of investing in more than one risky security. This is equivalent to the problem of choosing an
individual’s probability distribution of end-of-period wealth that is consistent with the

1 Ronald W. Masulis was the primary author of this chapter and has benefited from lecture notes on this topic by Herbert
Johnson.

From Chapter 4 of Financial Theory and Corporate Policy, Fourth Edition. Thomas E. Copeland, J. Fred Weston, Kuldeep

73
 State Preference Theory

Figure 1 Elementary π1 $10,000 State 1: You have the winning ticket

state-contingent claim.
Lottery ticket
π2
$0 State 2: You don't have the winning ticket

set of available risky securities and the individual’s initial wealth. The individual’s choice problem
is to find that portfolio or linear combination of risky securities that is optimal, given his or her
initial wealth and tastes. We assume a perfect capital market to ensure that there are no costs of
portfolio construction.

A. Uncertainty and Alternative Future States

Securities inherently have a time dimension. The securities investment decisions of individuals are
determined by their desired consumption over future time intervals. The passage of time involves
uncertainty about the future and hence about the future value of a security investment. From the
standpoint of the issuing firm and the individual investors, the uncertain future value of a security
can be represented as a vector of probable payoffs at some future date, and an individual’s portfolio
of investments is a matrix of possible payoffs on the different securities that compose the portfolio.
In the state preference model, uncertainty takes the form of not knowing what the state of
nature will be at some future date. To the investor a security is a set of possible payoffs, each
one associated with a mutually exclusive state of nature. Once the uncertain state of the world is
revealed, the payoff on the security is determined exactly. Thus a security represents a claim to a
vector (or bundle) of state-contingent payoffs.
In the simplest case, there are two possible outcomes with probabilities π1 and π2 and therefore
two mutually exclusive states of nature with probabilities π1 and π2. Take as an example an
investment in a lottery ticket with outcomes ($10,000, $0). With probability π1, state 1 is realized
and the lottery ticket pays off $10,000; with probability π2 = 1 − π1, state 2 is realized and the
lottery ticket pays off nothing (Fig. 1).
The probability of a state of nature occurring is thus equal to the probability of the associated
end-of-period security payoff. The states of nature are assumed to capture the fundamental causes
of economic uncertainty in the economy; for example, state 1 could represent peace and state 2
could represent war, or state 1 could represent prosperity and state 2 could represent depression.
Once the state of nature is known, the end-of-period payoff at each risky security is also known.
By summing over individual security holdings and then over individuals, it follows that once the
state of nature is known, individual and aggregate end-of-period wealth are also known.
In principle, there can be an infinite number of states of nature and thus an infinite number
of end-of-period payoffs for a risky asset. This set of states must meet the critical properties of
being mutually exclusive and exhaustive. That is to say, one and only one state of nature will be
realized at the end of the period, and the sum of the probabilities of the individual states of nature
equals one. It is also assumed that (1) individuals can associate an outcome from each security’s
probability distribution of its end-of-period payoff with each state of nature that could occur, and
(2) individuals are only concerned about the amount of wealth they will have if a given state occurs;

74
Edition
 State Preference Theory

Figure 2 States, pay- States of nature 1, 2, 3,..., S

offs, and securities. Prosperity
Normalcy
Recession
Depression Market securities j, k,..., J
Securities are

…
defined by patterns
of payoffs under
Payoffs different states
High
Medium
Low
Zero
…

once their wealth is known they are otherwise indifferent as to which state of nature occurs (i.e.,
individuals have state-independent utility functions).2

B. Definition of Pure Securities

Analytically, the generalization of the standard, timeless, microeconomic analysis under certainty
to a multiperiod economy under uncertainty with securities markets is facilitated by the concept
of a pure security. A pure or primitive security is defined as a security that pays $1 at the end of
the period if a given state occurs and nothing if any other state occurs. The concept of the pure
security allows the logical decomposition of market securities into portfolios of pure securities.3
Thus every market security may be considered a combination of various pure securities.
In terms of state preference theory, a security represents a position with regard to each possible
future state of nature. In Fig. 2, market securities are defined with respect to the characteristics of
their payoffs under each alternative future state. A market security thus consists of a set of payoff
characteristics distributed over states of nature. The complexity of the security may range from
numerous payoff characteristics in many states to no payoff at all in all but one state.

C. Complete Capital Market

In the state preference framework, uncertainty about securities’ future values is represented by
a set of possible state-contingent payoffs. Linear combinations of this set of state-contingent
security payoffs represent an individual’s opportunity set of state-contingent portfolio payoffs.
An important property of this opportunity set is determined by whether or not the capital market is
complete. When the number of unique linearly independent securities is equal to the total number
of alternative future states of nature, the market is said to be complete. For the case of three states of

2 For example, if an individual’s utility were a function of other individuals’ wealth positions as well as one’s own, then
the utility function would generally be state dependent.
3 Pure or primitive securities are often called Arrow-Debreu securities, since Arrow [1964] and Debreu [1959] set forth
their original specification.

75
Edition
 State Preference Theory

nature, suppose that a risk-free asset with payoff (1,1,1), an unemployment insurance contract with
payoff (1,0,0), and risky debt with payoff (0,1,1) all exist, but no other securities can be traded.
In this case we have three securities and three states of nature, but we do not have a complete
market since the payoff on the risk-free asset is just the sum of the payoffs on the other two market
securities; that is, the three securities are not linearly independent. If the market is incomplete, then
not every possible security payoff can be constructed from a portfolio of the existing securities.
For example, the security payoff (0,1,0) cannot be obtained from (1,1,1), (1,0,0), and (0,1,1). The
existing securities will, of course, have well-defined prices, but any possible new security not
spanned by these securities (i.e., cannot be created from the existing securities) will not have a
unique price.4
Suppose now that in addition to the security payoffs (1,1,1), (1,0,0), and (0,1,1), a stock
with payoff (0,1,3) also exists. Then among these four securities there are three that are linearly
independent state-contingent payoffs, and with three states the market is complete. Assuming the
market is perfect, any pattern of returns can be created in a complete market. In particular, a
complete set of pure securities with payoffs (1,0,0), (0,1,0), and (0,0,1) can be created as linear
combinations of existing securities. It takes some linear algebra to figure out how to obtain the
pure securities from any arbitrary complete set of market securities, but once we know how to form
them, it is easy to replicate any other security from a linear combination of the pure securities. For
example: a security with a payoff (a,b,c) can be replicated by buying (or short selling if a, b, or c
is negative) a of (1,0,0), b of (0,1,0), and c of (0,0,1).5
Given a complete securities market, we could theoretically reduce the uncertainty about our
future wealth to zero. It does not make any difference which uncertain future state of nature will
actually occur. That is, by dividing our wealth in a particular way among the available securities,
we could, if we chose, construct a portfolio that was equivalent to holding equal amounts of all the
pure securities. This portfolio would have the same payoff in every state even though the payoffs
of individual securities varied over states.6
Without going through a complex solution process to attain the general equilibrium results that
the concept of a pure security facilitates, we shall convey the role of the concept of a pure security
in a more limited setting. We shall demonstrate how in a perfect and complete capital market the
implicit price of a pure security can be derived from the prices of existing market securities and
how the prices of other securities can then be developed from the implicit prices of pure securities.

D. Derivation of Pure Security Prices

Given that we know the state-contingent payoff vectors of both the market securities and the pure
securities, we wish to develop the relationship between the prices of the market securities and pure
securities in a perfect and complete capital market.

4 One person might think the security with payoff (0,1,0) is worth more than someone else does, but if the security cannot
be formed from a portfolio of existing market securities, then these virtual prices that different people would assign to this
hypothetical security need not be the same.
5 See Appendix A for a general method of determining whether a complete market exists.
6 While a complete market may appear to require an unreasonably large number of independent securities, Ross [1976]
showed that in general if option contracts can be written on market securities and market securities have sufficiently variable
payoffs across states, an infinite number of linearly independent security and option payoffs can be formed from a small
number of securities.

76
Edition
 State Preference Theory

Table 1 Payoffs in Relation to Prices of Baskets of Fruit

Bananas Apples Prices ∗
Basket 1 10 20 $8
Basket 2 30 10 $9
* The probabilities of the states are implicit in the prices.

The following notation will be used throughout this chapter:

ps = prices of pure securities,

pj = prices of market securities,
πs = state probabilities—individuals’ beliefs about the relative
likelihoods of states occurring,
Qs = number of pure securities.

Let us begin with an analogy. The Mistinback Company sells baskets of fruit, limiting its sales
to only two types of baskets. Basket 1 is composed of 10 bananas and 20 apples and sells for
$8. Basket 2 is composed of 30 bananas and 10 apples and sells for $9. The situation may be
summarized by the payoffs set forth in Table 1.
Using the relationships in Table 1, we can solve for the prices of apples and bananas separately.
Let us denote apples by A, bananas by B, the baskets of fruit by 1 and 2, and the quantity of apples
and bananas in a basket by Qj A and Qj B , respectively. Using this notation, we can express the
prices of the two baskets as follows:

p1 = pAQ1A + pB Q1B , p2 = pAQ2A + pB Q2B .

Only pA and pB are unknown. Thus there are two equations and two unknowns, and the system
is solvable as follows (substitute the known values in each equation):

$8 = pA20 + pB 10, (a) $9 = pA10 + pB 30. (b)

Subtract three times Eq. (a) from Eq. (b) to obtain pA:

$9 = pA10 + pB 30
−$24 = −pA60 − pB 30
−$15 = −pA50
pA = $.30

Then substituting the value pA into Eq. (a), we have

$8 = ($.30)20 + pB 10 = $6 + pB 10
$2 = pB 10
pB = $.20.

Given that we know the prices of the market securities, we may now apply this same analysis
to the problem of determining the implicit prices of the pure securities. Consider security j , which

77
Edition
 State Preference Theory

Table 2 Payoff Table for Securities j and k

Security State 1 State 2
j $10 $20 pj = $8
k $30 $10 pk = $9

pays $10 if state 1 occurs and $20 if state 2 occurs; its price is $9. Note that state 1 might be
a gross national product (GNP) growth of 8% in real terms during the year, whereas state 2
might represent a GNP growth rate of only 1% in real terms. This information is summarized
in Table 2.
Any individual security is similar to a mixed basket of goods with regard to alternative future
states of nature. Recall that a pure security pays $1 if a specified state occurs and nothing if any
other state occurs. We may proceed to determine the price of a pure security in a matter analogous
to that employed for the fruit baskets.
The equations for determining the price for two pure securities related to the situation described
are
p1Qj 1 + p2 Qj 2 = pj ,

p1Qk1 + p2 Qk2 = pk,

where Qj 1 represents the quantity of pure securities paying $1 in state 1 included in security j .
Proceeding analogously to the situation for the fruit baskets, we insert values into the two equations.
Substituting the respective payoffs for securities j and k, we obtain $.20 as the price of pure secu-
rity 1 and $.30 as the price of pure security 2:

p110 + p2 20 = $8,
p130 + p2 10 = $9,
p1 = $.20, p2 = $.30.

It should be emphasized that the p1 of $.20 and the p2 of $.30 are the prices of the two pure
securities and not the prices of the market securities j and k. Securities j and k represent portfolios
of pure securities. Any actual security provides different payoffs for different future states. But
under appropriately defined conditions, the prices of market securities permit us to determine the
prices of pure securities. Thus our results indicate that for pure security 1 a $.20 payment is required
for a promise of a payoff of $1 if state 1 occurs and nothing if any other states occur. The concept
of pure security is useful for analytical purposes as well as for providing a simple description of
uncertainty for financial analysis.

E. No-Arbitrage Profit Condition

Capital market equilibrium requires that market prices be set so that supply equals demand for
each individual security. In the context of the state preference framework, one condition necessary
for market equilibrium requires that any two securities or portfolios with the same state-contingent

78
Edition
 State Preference Theory

payoff vectors must be priced identically.7 Otherwise, everyone would want to buy the security
or portfolio with the lower price and to sell the security or portfolio with the higher price. If both
securities or portfolios are in positive supply, such prices cannot represent an equilibrium. This
condition is called the single-price law of markets.
If short selling is allowed in the capital market, we can obtain a second related necessary
condition for market equilibrium—the absence of any riskless arbitrage profit opportunity. To
short-sell a security, an individual borrows the security from a current owner and then immediately
sells the security in the capital market at the current price. Then, at a later date, the individual
goes back to the capital market and repurchases the security at the then-current market price and
immediately returns the security to the lender. If the security price falls over the period of the short
sale, the individual makes a profit; if the security price rises, he or she takes a loss. In either case
the short seller’s gain or loss is always the negative of the owner’s gain or loss over this same
period.
When two portfolios, A and B, sell at different prices, where pA > pB , but have identical state-
contingent payoff vectors, we could short-sell the more expensive portfolio and realize a cash flow
of pA, then buy the less expensive portfolio, for a negative cash flow of pB . We would realize
a positive net cash flow of (pA − pB ), and at the end of the period, we could at no risk take our
payoff from owning portfolio B to exactly repay our short position in portfolio A. Thus the positive
net cash flow at the beginning of the period represents a riskless arbitrage profit opportunity. Since
all investors are assumed to prefer more wealth to less, this arbitrage opportunity is inconsistent
with market equilibrium.
In a perfect and complete capital market, any market security’s payoff vector can be exactly
replicated by a portfolio of pure securities. Thus, it follows that when short selling is allowed, the
no-arbitrage profit condition requires that the price of the market security be equal to the price of
any linear combination of pure securities that replicates the market security’s payoff vector.

F. Economic Determinants of Security Prices

To gain an understanding of what determines the price of a market security, we will first consider
what determines the price of individual pure securities. Since a market security can always be
constructed from the set of pure securities in a complete market, we can then answer the first
question as well.
The prices of the pure securities will be determined by trading among individuals. Even if these
pure securities themselves are not directly traded, we can still infer prices for them in a complete
market from the prices of the market securities that are traded. The prices of pure securities will
be shown to depend on
1. time preferences for consumption and the productivity of capital;
2. expectations as to the probability that a particular state will occur;
3. individuals’ attitudes toward risk, given the variability across states of aggregate end-of-period
wealth.

7 Thiscondition implies the absence of any first-order stochastically dominated market securities. Otherwise the former
payoff per dollar of investment would exceed the latter payoff per dollar of investment in every state. The latter security
would be first-order stochastically dominated by the former security.

79
Edition
 State Preference Theory

To understand how time preferences and the productivity of capital affect security prices, we
need to recognize that a riskless security can always be constructed in a complete capital market
simply by forming a portfolio composed of one pure security for each state. The payoff on this
portfolio is riskless since a dollar will be paid regardless of what state is realized. In the case of
three states the price of this riskless portfolio is the sum of the prices of the three individual pure
securities (e.g., p1 + p + p3 = .8). The price of a riskless claim to a dollar at the end of the period
 is
just the present value of a dollar discounted at the risk-free rate rf , that is to say, 1/(1 + rf ) = ps .
If there is a positive time value of money, the riskless interest rate will be positive. The actual size
of this interest rate will reflect individual time preferences for consumption and the productivity
of capital, just as is the case in a simple world of certainty.8 Thus one determinant of the price of a
pure security paying a dollar if state s occurs is the market discounted rate on certain end-of-period
dollar payoff.
The second determinant of a pure security’s price, and a cause for differences in security
prices, is individuals’ beliefs concerning the relative likelihood of different states occurring. These
beliefs are often termed state probabilities, πs . Individuals’ subjective beliefs concerning state
probabilities can differ in principle. However, the simplest case is one in which individuals agree on
the relative likelihoods of states. This assumption is termed homogeneous expectations and implies
that there is a well-defined set of state probabilities known to all individuals in the capital market.
Under the assumption of homogeneous expectations, the price of a pure (state-contingent) security,
ps , can be decomposed into the probability of the state, πs , and the price, θs , of an expected dollar
payoff contingent on state s occurring, ps = πs · θs . This follows from the fact that pure security s
pays a dollar only when s is realized. Thus the expected end-of-period payoff on pure security s is
a dollar multiplied by the probability of state s occurring. This implies that we can decompose the
end-of-period expected payoff into an expected payoff of a dollar and the probability of state s.
Even when prices contingent on a particular state s occurring are the same across states (θs = θt ,
for all s and t), the prices of pure securities will differ as long as the probabilities of states occurring
are not all identical (πs  = πt , for all s and t).
A useful alternative way to see this point is to recognize that the price of a pure security is equal
to its expected end-of-period payoff discounted to the present at its expected rate of return:

$1 · πs
ps = ,
1 + E(Rs )

where 0 < ps < 1. Thus the pure security’s expected rate of return is

$1 · πs $1
E(Rs ) = − 1= − 1, where 0 < θs < 1,
ps θs

since ps = πs θs under the assumption of homogeneous expectations. So if the θs ’s were identical

across states, the expected states of return would be equal for all pure securities. But given that

8 The property that individuals prefer to consume a dollar of resources today, rather than consume the same dollar of
resources tomorrow, is called the time preference for consumption. An individual’s marginal rate of time preference for
consumption is equal to his or her marginal rate of substitution of current consumption and certain end-of-period consumption.
In a perfect capital market, it was also shown that the marginal rates of time preference for all individuals are equal to the
market interest rate.

80
Edition
 State Preference Theory

the probabilities across states differ, the expected payoffs across pure securities must also differ.
If expected payoffs vary, expected rates of return can be the same only when the prices of the pure
securities vary proportionally with the state probabilities.
The third determinant of security prices, and a second cause for differences in these prices,
is individuals’ attitudes toward risk when there is variability in aggregate wealth across states.
Assuming that individuals are risk averse, they will diversify by investing in some of each pure
security to ensure that they are not penniless regardless of what state is realized.9 In fact, if the
prices, θs ’s, of expected payoffs of a dollar contingent on a particular state occurring were the same
for all states (and thus the expected rates of return of pure securities are all equal), then each risk-
averse individual would want to invest in an equal number of each pure security so as to eliminate all
uncertainty about his or her future wealth. Not everyone can do this, however, in the economy, and
it must be borne by someone. Consider the following example. End-of-period aggregate wealth can
be one, two, or three trillion dollars, depending on whether the depressed, normal, or prosperous
state occurs; then the average investor must hold a portfolio with a payoff vector of the form
(X, 2X, 3X). Because individuals are risk averse, dollar payoffs are more valuable in states where
they have relatively low wealth, which in this example is state 1. In order for people to be induced to
bear the risk associated with a payoff vector of the form (X, 2X, 3X), pure security prices must be
adjusted to make the state 1 security relatively expensive and the state 3 security relatively cheap.
In other words, to increase demand for the relatively abundant state 3 securities, prices must adjust
to lower the expected rate of return on state 1 securities and to raise the expected rate of return on
state 3 securities.
If aggregate wealth were the same in some states, then risk-averse investors would want
to hold the same number of pure securities for these states and there would be no reason for
prices of expected dollar payoffs to be different in these states. Investors would not want to hold
unequal numbers of claims to the states with the same aggregate wealth because this would mean
bearing risk that could be diversified away, and there is no reason to expect a reward for bearing
diversifiable risk. So it is the prospect of a higher portfolio expected return that induces the risk-
averse investors to bear nondiversifiable risk. Thus risk aversion combined with variability in
end-of-period aggregate wealth causes variation in the prices (θs ’s) of dollar expected payoffs
across states, negatively related to the aggregate end-of-period wealth or aggregate payoffs across
states. This in turn causes like variations in the pure security prices.
There is a very important condition implicit in the previous discussion. We found that when
investors are risk averse, securities that pay off relatively more in states with low aggregate wealth
have relatively low expected rates of return, whereas securities that pay off relatively more in states
with high aggregate wealth have relatively high expected rates of return. Since aggregate wealth is
equal to the sum of the payoffs on all market securities, it is also termed the payoff on the market
portfolio. Securities with state-contingent payoffs positively related to the state-contingent payoffs
on the market portfolio, and which therefore involve significant nondiversifiable risk bearing, have
higher expected rates of return than securities that have payoffs negatively or less positively related
to the payoffs on the market portfolio, and which therefore involve little diversifiable risk bearing.
It follows from this analysis that a pure security price can be decomposed into three factors:

9 This also requires the utility function to exhibit infinite marginal utility at a zero wealth level.

81
Edition
 State Preference Theory

   
$1πs $1 1 + rf
ps = πs θs = = πs
1 + E(Rs ) 1 + rf 1 + E(Rs )
   
$1 E(Rs ) − rf
= πs 1 − , where E(Rs ) ≥ rf .
1 + rf 1 + E(Rs )

The first factor is an end-of-period dollar payoff discounted to the present at the riskless rate. It
is multiplied by the second factor, which is the probability of payoff. The third factor is a risk
adjustment factor. Note that if investors are all risk neutral, the expected rate of return on all
securities will be equal to the riskless interest rate, in which case the above risk adjustment factor
(i.e., the third factor) becomes one. In summary, security prices are affected by (1) the time value
of money, (2) the probability beliefs about state-contingent payoffs, and (3) individual preferences
toward risk and the level of variability in aggregate state-contingent payoffs or wealth (i.e., the
level of nondiversifiable risk in the economy).

G. Optimal Portfolio Decisions

Now that we have developed the basic structure of state preference theory, we will return to the
problem of optimal portfolio choice in a perfect and complete capital market. This will then be
followed by an analysis of a firm’s optimal investment problem, also in a perfect and complete
capital market. Since any portfolio payoff pattern can be constructed from the existing market
securities or from a full set of pure securities in a complete capital market, we can obtain the same
optimal portfolio position whether we frame the analysis in terms of market securities or pure
securities. Since pure securities are much simpler to analyze, we will phrase the optimal portfolio
problem in terms of these securities. Thus we can write an individual’s expected utility of end-
of-period wealth as πs U (Qs ), where Qs = number of pure securities paying a dollar if state s
occurs. In this context, Qs represents the number of state s pure securities the individual buys as
well as his or her end-of-period wealth if state s occurs.
Now consider the problem we face when we must decide how much of our initial wealth, W0,
to spend for current consumption, C, and what portfolio of securities to hold for the future. We
wish to solve the problem10
 

max u(C) + πs U (Qs ) (1)
s

subject to

ps Qs + $1C = W0 . (2)
s

That is, we are maximizing our expected utility of current and future consumption (Eq. 1)
subject to our wealth constraint (Eq. 2). Our portfolio decision consists of the choices we make

10 This formulation assumes that the utility function is separable into utility of current consumption and utility of end-of-
period consumption. In principle, the utility functions u(C) and U (Qs ) can be different functions.

82
Edition
 State Preference Theory

for Qs , the number of pure securities we buy for each state s. Note that there is no explicit
discounting of future utility, but any such discounting could be absorbed in the functional form for
U (Qs ). In addition, the ps ’s include an implicit market discount rate. There is no need to take an
expectation over u(C), our utility of current consumption, since there is no uncertainty concerning
the present.
There are two ways to maximize expected utility subject to a wealth constraint. We could solve
(2) for one of the Qs ’s, say Q1, and then eliminate this variable from (1). Sometimes this is the
easiest way, but more often it is easier to use the Lagrange multiplier method:


 
L = u(C) + πs U (Qs ) − λ ps Qs + $1C − W0 , (3)
s s

where λ is called a Lagrange multiplier. The Lagrange multiplier λ is a measure of how much
our utility would increase if our initial wealth were increased by $1 (i.e., the shadow price for
relaxing the constraint). To obtain the investor’s optimal choice of C and Qs ’s, we take the partial
derivatives with respect to each of these variables and set them equal to zero. Taking the partial
derivative with respect to C yields

∂L
= u(C) − $1λ = 0, (4)
∂C

where the prime denotes partial differentiation with respect to the argument of the function. Next,
we take partial derivatives with respect to Q1, Q2, and so on. For each Qt , we will pick up one
term from the expected utility and one from the wealth constraint (all other terms vanish):

∂L
= πt U (Qt ) − λpt = 0, (5)
∂Q1

where πt U (Qt ) = expected marginal utility of an investment Qt in pure security s. We also take
the partial derivative with respect to λ:

∂L 
= ps Qs + $1C − W0 = 0. (6)
∂λ s

This just gives us back the wealth constraint. These first-order conditions allow us to determine
the individual’s optimal consumption/investment choices.11
As an example, consider an investor with a logarithmic utility function of wealth and initial
wealth of $10,000. Assume a two-state world where the pure security prices are .4 and .6 and the
state probabilities are 31 and 23 , respectively. The Lagrangian function is

1 2
L = ln C + ln Q1 + ln Q2 − λ(.4Q1 + .6Q2 + C − 10,000),
3 3

and the first-order conditions are

11 We are also assuming that the second-order conditions for a maximum hold.

83
Edition
 State Preference Theory

∂L 1 1
= − λ = 0, which implies C= , (a)
∂C C λ
∂L 1 1
= − .4λ = 0, which implies Q1 = , (b)
∂Q1 3Q1 1.2λ
∂L 2 1
= − .6λ = 0, which implies Q2 = , (c)
∂Q1 3Q2 .9λ
∂L
= 10,000 − C − .4Q1 − .6Q2 = 0. (d)
∂λ

Substituting Eqs. (a), (b), and (c) into (d) yields

1 .4 .6
+ + = 10,000, (d)
λ 1.2λ .9λ

and multiplying by λ yields

1 2 1
1+ + = 10,000λ, which yields λ = . (d)
3 3 5,000

Now, substituting this value of λ back into Eqs. (a), (b), and (c) yields the optimal consumption and
investment choices, C = $5,000, Q1 = 4,166.7, and Q2 = 5,555.5. Substituting these quantities
back into the wealth constraint verifies that this is indeed a feasible solution. The investor in this
problem divides his or her wealth equally between current and future  consumption, which is what
we should expect since the risk-free interest rate is zero—that is, ps = 1 = 1/(1 + r)—and there
is no time preference for consumption in this logarithmic utility function. However, the investor
does buy more of the state 2 pure security since the expected rate of return on the state 2 pure
security is greater. Because the utility function exhibits risk aversion, the investor also invests
some of his or her wealth in the state 1 pure security.
In this example we assumed that the investor is a price taker. In a general equilibrium frame-
work, the prices of the pure securities would be determined as part of the problem; that is, they
would be endogenous. The prices would be determined as a result of the individuals’ constrained
expected utility maximization (which determines the aggregate demands for securities). The criti-
cal condition required for equilibrium is that the supply of each market security equal its aggregate
demand. In a complete capital market this equilibrium condition can be restated by saying that the
aggregate supply of each pure security is equal to its aggregate demand.

H. T(and
he Efficient Set with Two Risky Assets
No Risk-Free Asset)
In a complete capital market, we can obtain a number of important portfolio optimality conditions.
These conditions hold for any risk-averse expected utility maximizer. Rewriting Eq. (4) and
Eq. (5) in terms of λ and eliminating λ yields two sets of portfolio optimality conditions:

πt U (Qt ) pt
= for any state t (7)
u(C) $1

84
Edition
 State Preference Theory

Figure 3 Optimal Qt
consumption/investment
decisions. Indifference curve

πtU′(Qt )
Slope = – ———–
u′(C )

Market
line

Slope = –pt /$1

C

and

πt U (Qt ) p
= t for any two states s and t. (8)
πs U (Qs ) ps

In both cases, the optimal allocation of wealth represents choosing C and the Qs ’s so that the ratio
of expected marginal utilities equals the ratio of market prices for the C and the Qs ’s. That is, the
optimal consumption and investment choices involve choosing points on the various indifference
curves (curves of constant expected utility) that are tangent to the associated market lines. This is
equivalent to choosing consumption and investment weights so that the slopes of the indifference
curves (which are defined as the negative of the marginal rates of substitution) representing current
consumption and future consumption contingent on state t (as in Fig. 3) or representing future
consumption contingent on state t (as in Fig. 4) are equal to the slopes of the respective market
lines (representing the market exchange rates, e.g., −pt /ps ).
An alternative way of stating the optimality conditions of the above portfolio is that the expected
marginal utilities of wealth in state s, divided by the price of the state s pure security, should be equal
across all states, and this ratio should also be equal to the marginal utility of current consumption.
This is a reasonable result; if expected marginal utility per pure security price were high in one
state and low in another, then we must not have maximized expected utility. We should increase
investment in the high expected marginal utility security at the expense of the security yielding
low expected marginal utility. But as we do that, we lower expected marginal utility where it is
high and raise it where it is low, because a risk-averse investor’s marginal utility decreases with
wealth (his or her utility function has a positive but decreasing slope). Finally, when Eq. (8) is
satisfied, there is no way left to provide a further increase of expected utility.12
When investors’ portfolio choices over risky securities are independent of their individual
wealth positions, we have a condition known as portfolio separation. This condition requires
that there are either additional restrictions on investor preferences or additional restrictions on

12 This entire procedure will ordinarily not work if the investor is risk neutral instead of risk averse. A risk-neutral investor
will plunge entirely into the security with the highest expected return. He or she would like to invest more in this security,
but, being already fully invested in it, cannot do so. Equation (8) will not hold for risk neutrality.

85
Edition
 State Preference Theory

Figure 4 Optimal Qt
portfolio decisions. Indifference curve

πtU′(Qt )
Slope = – ———–
πsU′(Qs)

Market
line

Slope = –pt /ps

Qs

security return distributions.13 Under portfolio separation, investors choose among only a few
basic portfolios of market securities. Thus the importance of having a complete market is greatly
decreased. Recall that when capital markets are incomplete, individuals are limited in their choices
of state-contingent payoff patterns to those payoff patterns that can be constructed as linear
combinations of existing market securities. However, with portfolio separation, investors will often
find that the infeasible payoff opportunities would not have been chosen even if they were available.
Thus under portfolio separation, investor portfolio decisions will often be unaffected by whether
or not the capital market is complete.
Portfolio separation has been shown to depend on the form of the utility function of individuals
and the form of the security return distributions. In the special case where investor utility functions
of wealth are quadratic, or security returns are joint-normally distributed, portfolio separation
obtains. With the addition of homogeneous expectations, portfolio separation provides sufficient
conditions for a security-pricing equation. This security-pricing relationship can be expressed in
terms of means and variances and is called the capital asset pricing model. The resulting form of
the security-pricing equation is particularly convenient for formulating testable propositions and
conducting empirical studies. Many of the implications of portfolio separation in capital markets
appear to be consistent with observed behavior and have been supported by empirical tests.

13 Cass and Stiglitz [1970] proved that for arbitrary security return distributions, utility functions with the property of
linear risk tolerance yield portfolio separation. The risk tolerance of the utility function is the reciprocal of the Pratt-Arrow
measure of the absolute risk aversion. Thus a linear risk-tolerance utility function can be expressed as a linear function of
wealth:
−u(W )/U  (W ) = a + bW. (9)
If investors also have homogeneous expectations about state probabilities and all investors have the same b, then there
is two-fund separation, where all investors hold combinations of two basic portfolios.
Utility functions exhibiting linear risk tolerance include the quadratic, logarithmic, power, and exponential functions.
Ross [1976] proved that for arbitrary risk-averse utility functions a number of classes of security return distributions
(including the normal distribution, some stable Paretian distributions, and some distributions that are not stable Paretian,
e.g., fat-tailed distributions with relatively more extreme values) yield portfolio separation.

86
Edition
 State Preference Theory

I. Firm Valuation, the Fisher Separation Principle,

and Optimal Investment Decisions14
In state preference theory, individuals save when they purchase a firm’s securities. Firms obtain
resources for investment by issuing securities. Securities are precisely defined as conditional
or unconditional payoffs in terms of alternative future states of nature. All individual and firm
decisions are made, and all trading occurs at the beginning of the period. Consumers maximize
their expected utility of present and future consumption and are characterized by their initial
endowments (wealth) and their preferences. Firms are characterized by production functions that
define the ability to transform current resources into state-contingent future consumption goods; for
example, where Ij is the initial investment, Qsj = φ(Ij , s). Total state-contingent output produced
by a firm must equal the sum of the payoffs from all securities issued by a firm.
A firm maximizes an objective function that, in its most general form, is maximization
of the expected utility of its current shareholders. To do this it may appear that a firm would
need to know the utility functions of all its current shareholders. However, in a perfect capital
market (a frictionless and perfectly competitive market) under certainty, actions that maximize
the price of the firm’s shares maximize both the wealth and the utility of each current
shareholder. So managers need only know the market discount rate and the cash flows of
their investment projects to make optimal investment decisions. This separation of investment/
operating decisions of firms from shareholder preferences or tastes is termed the Fisher
separation principle.
In shifting from a firm’s decision making under certainty to a world with uncertainty, it
is important to know under what conditions, if any, Fisher separation continues to hold. It
can be shown that firms that are maximizing the price of current shares are also maximizing
current shareholders’ expected utility when the capital market is (1) perfectly competitive
and frictionless and (2) complete. The first condition ensures that one firm’s actions will
not be perceived to affect other firms’ market security prices, whereas the second ensures
that the state space “spanned” by the existing set of linearly independent market securities
(i.e., the set of risky opportunities) is unaffected by the firm’s actions. Thus the firm’s actions
affect shareholders’ expected utility only by affecting their wealth through changes in the
firm’s current share price.
The two conditions of a perfect and complete capital market ensure that the prices of a full set
of pure securities can be obtained from the prices of the market securities, and vice versa, given
the state-contingent payoff vectors of the market securities. As a result, the firm’s objective
function to maximize current share price can be phrased in terms of a portfolio
 of pure securities
that replicates its shares. The firm’s objective function then becomes Qj s ps , where Qj s
is defined as the state s–contingent end-of-period payoff on firm j ’s existing securities. In
this formulation, the price of a firm’s current shares is determined by (1) the firm’s state-
contingent production function Qj s = φj (Ij , s), which transforms current resources into state-
contingent future payoffs, and (2) the initial investment Ij , which represents the current cost to the
firm of producing its state-contingent payoff. It follows that the price Yj , for which the current

14 Hirshleifer [1964, 1965, 1966] and Myers [1968] were among the first papers to apply state preference theory to corporate
finance problems.

87
Edition
 State Preference Theory

Table 3 Firm A’s Stock and Investment Project Payoffs

State-Contingent Payoff on
States of Nature Stock Payoff Firm A’s Proposed Investment Project
State 1 100 10
State 2 30 12
Firm A’s stock price = 62; initial investment cost of its project = 10.

Table 4 Firm B’s Stock and Investment Project Payoffs

State-Contingent Payoff on
States of Nature Stock Payoff Firm B’s Proposed Investment Project
State 1 40 12
State 2 90 6
Firm B’s stock price = 56; initial investment cost of its project = 8.

owners could sell the firm prior to making the investment Ij , is


Yj = ps Qj s − Ij . (10)
s

For indivisible investment projects with finite scale, the optimal investment rule is to accept all
projects with positive net present value. In this context, Eq. (10) represents the net present value
of the project’s state-contingent net cash flow.
It is important to note that acceptance of positive NPV investments increases the price of the
firm’s current stock and therefore the wealth and expected utility of all current shareholders in a
perfect and complete capital market. Since all shareholders are made better off by these investment
decisions, these firm investment decisions are unanimously supported by all the firm’s current
shareholders. However, if the capital market is incomplete or imperfect, this is not necessarily true,
because the firm’s investment decisions may affect the price of other firms’ shares or the feasible
set of state-contingent payoffs. As a result, increasing the price of a firm’s shares may not increase
the wealth of all current shareholders (since the prices of some of their other shareholdings may
fall) and may not maximize shareholder expected utility (since the opportunity set of feasible end-
of-period payoffs may have changed).15
Let us now consider an example of a firm investment decision problem in a two-state world of
uncertainty. Assume that all investors are expected utility maximizers and exhibit positive marginal
utility of wealth (i.e., more wealth is preferred to less). Consider the following two firms and their
proposed investment projects described in Tables 3 and 4.
To determine whether either firm should undertake its proposed project, we need to first
determine whether the capital market is complete. Since the state-contingent payoffs of the two
firms’ stocks are linearly independent, the capital market is complete. In a complete market, the
Fisher separation principle holds, so that the firm need only maximize the price of current shares

15 See DeAngelo [1981] for a critical analysis of the unanimity literature and a careful formulation of the conditions under
which it holds in incomplete and complete capital markets.

88
Edition
 State Preference Theory

to maximize its current shareholders’ expected utility. This requires that the firm invest only in
positive net present value investment projects, which requires knowing the pure security prices in
the two states. These two prices can be obtained from the market prices of the two firms’ stocks
and their state-contingent payoffs by solving the two simultaneous equations

100p1 + 30p2 = 62,

40p1 + 90p2 = 56,

to obtain the solution p1 = .5 and p2 = .4. To calculate the net present value of the two projects,
we use the NPV definition in (10):

NPVA = 10p1 + 12p2 − I0A = 10(.5) + 12(.4) − 10 = −.2

and

NPVB = 12p1 + 6p2 − I0B = 12(.5) + 6(.4) − 8 = .4.

Since firm A’s project has a negative NPV, it should be rejected, whereas firm B’s project should
be accepted since it has a positive NPV.
In examining this optimal investment rule, it should be clear that the prices of the pure securities
affect the firm’s investment decisions. It follows that since these security prices are affected by
(1) time preference for consumption and the productivity of capital, (2) the probability of state-
contingent payoffs, and (3) individual preferences toward risk and the level of nondiversifiable risk
in the economy, firm investment decisions are also affected by these factors.
We have applied state preference theory to the firm’s optimal investment decision while assum-
ing that the firm has a simple capital structure represented by shares of stock. However, it is also
possible to allow the firm to have a more complicated capital structure, which may include various
debt, preferred stock, and warrant contracts. In doing this, state preference theory can be used to
address the important question of a firm’s optimal financing decision.16 For this purpose it has been
found useful to order the payoffs under alternative states. One can think of the payoffs for future
states as arranged in an ordered sequence from the lowest to the highest payoff. Keeping in mind
the ordered payoffs for alternative future states, we can specify the conditions under which a se-
curity such as corporate debt will be risk free or risky.17
The state preference model has also been very useful in developing option pricing theory. By
combining securities with claims on various portions of the ordered payoffs and by combining long
and short positions, portfolios with an infinite variety of payoff characteristics can be created. From
such portfolios various propositions with regard to option pricing relationships can be developed.18

16 There are many examples of the usefulness of state-preference theory in the area of optimal capital structure or financing
decisions; see, for example, Stiglitz [1969], Mossin [1977], and DeAngelo and Masulis [1980a, 1980b].
17 For applications of this approach, see Kraus and Litzenberger [1973] and DeAngelo and Masulis [1980a].
18 For some further applications of state preference theory to option pricing theory, see Merton [1973] and Ross [1976],
and for application to investment decision making, see Appendix B at the end of this chapter.

89
Edition
 State Preference Theory

Summary
Wealth is held over periods of time, and the different future states of nature will change the
value of a person’s wealth position over time. Securities represent positions with regard to the
relation between present and future wealth. Since securities involve taking a position over time,
they inherently involve risk and uncertainty.
The states of nature capture a wide variety of factors that influence the future values of risky
assets. Individuals must formulate judgements about payoffs under alternative future states of
nature. From these state-contingent payoffs and market prices of securities, the prices of the
underlying pure securities can be developed in a complete and perfect capital market. Given these
pure security prices, the price of any other security can be determined from its state-contingent
payoff vector. Conceptually, the equilibrium prices of the pure securities reflect the aggregate risk
preferences of investors and investment opportunities of firms. Furthermore, the concept of a pure
security facilitates analytical solutions to individuals’ consumption/portfolio investment decisions
under uncertainty.
The state preference approach is a useful way of looking at firm investment decisions under
uncertainty. In a perfect and complete capital market the net present value rule was shown
to be an optimal firm investment decision rule. The property that firm decisions can be made
independently of shareholder utility functions is termed the Fisher separation principle. State
preference theory also provides a conceptual basis for developing models for analyzing firm
capital structure decisions and the pricing of option contracts. Thus the state preference approach
provides a useful way of thinking about finance problems both for the individual investor and for
the corporate manager.
The state-preference model has been shown to be very useful in a world of uncertainty. It can
be used to develop optimal portfolio decisions for individuals and optimal investment rules for
firms. We have found that in perfect and complete capital markets a set of equilibrium prices
of all outstanding market securities can be derived. Further, these prices have been shown to be
determined by (1) individual time preferences for consumption and the investment opportunities
of firms, (2) probability beliefs concerning state-contingent payoffs, and (3) individual preferences
toward risk and the level of nondiversifiable risk in the economy.

PROBLEM SET
1 Security A pays $30 if state 1 occurs and $10 if state 2 occurs. Security B pays $20 if state 1
occurs and $0 if state 2 occurs. The price of security A is $5, and the price of security B is $10.
(a) Set up the payoff table for securities A and B.
(b) Determine the prices of the two pure securities.

90
Edition
 State Preference Theory

2 You are given the following information:

Payoff
Security State 1 State 2 Security Prices
j $12 $20 pj = $22
k 24 10 pk = 20

(a) What are the prices of pure security 1 and pure security 2?
(b) What is the initial price of a third security i, for which the payoff in state 1 is $6 and
the payoff in state 2 is $10?
3 Interplanetary starship captain Josė Ching has been pondering the investment of his recent
pilot’s bonus of 1,000 stenglers. His choice is restricted to two securities: Galactic Steel, selling
for 20 stenglers per share, and Nova Nutrients, at 10 stenglers per share. The future state of his
solar system is uncertain. If there is a war with a nearby group of asteroids, Captain Ching expects
Galactic Steel to be worth 36 stenglers per share. However, if peace prevails, Galactic Steel will
be worth only 4 stenglers per share. Nova Nutrients should sell at a future price of 6 stenglers per
share in either eventuality.
(a) Construct the payoff table that summarizes the starship captain’s assessment of future
security prices, given the two possible future states of the solar system. What are the prices
of the pure securities implicit in the payoff table?
(b) If the captain buys only Nova Nutrients shares, how many can he buy? If he buys only
Galactic Steel, how many shares can he buy? What would be his final wealth in both cases
in peace? At war?
(c) Suppose Captain Ching can issue (sell short) securities as well as buy them, but he must
be able to meet all claims in the future. What is the maximum number of Nova Nutrients
shares he could sell short to buy Galactic Steel? How many shares of Galactic Steel could
he sell short to buy Nova Nutrients? What would be his final wealth in both cases and in
each possible future state?
(d) Suppose a third security, Astro Ammo, is available and should be worth 28 stenglers
per share if peace continues and 36 stenglers per share if war breaks out. What would be
the current price of Astro Ammo?
(e) Summarize the results of (a) through (d) on a graph with axes W1 and W2.
(f) Suppose the captain’s utility function can be written as U = W1.8W2.2. If his investment
is restricted to Galactic Steel and/or Nova Nutrients, what is his optimal portfolio (i.e., how
many shares of each security should he buy or sell)?
4 Ms. Mary Kelley has initial wealth W0 = $1,200 and faces an uncertain future that she
partitions into two states, s = 1 and s = 2. She can invest in two securities, j and k, with initial
prices of pj = $10 and pk = $12, and the following payoff table:

Payoff
Security s=1 s=2
j $10 $12
k 20 8

91
Edition
 State Preference Theory

(a) If she buys only security j , how many shares can she buy? If she buys only security k,
how many can she buy? What would her final wealth, Ws , be in both cases and each state?
(b) Suppose Ms. Kelley can issue as well as buy securities; however, she must be able to
meet all claims under the occurrence of either state. What is the maximum number of shares
of security j she could sell to buy security k? What is the maximum number of shares of
security k she could sell to buy security j ? What would her final wealth be in both cases
and in each state?
(c) What are the prices of the pure securities implicit in the payoff table?
(d) What is the initial price of a third security i for which Qi1 = $5 and Qi2 = $12?
(e) Summarize the results of (a) through (d) on a graph with axes W1 and W2.
(f) Suppose Ms. Kelley has a utility function of the form U = W1.6W2.4. Find the optimal
portfolio, assuming the issuance of securities is possible, if she restricts herself to a portfolio
consisting only of j and k. How do you interpret your results?
5 Two securities have the following payoffs in two equally likely states of nature at the end of
one year:

Payoff
Security s=1 s=2
j $10 $20
k 30 10

Security j costs $8 today, whereas k costs $9, and your total wealth is currently $720.
(a) If you wanted to buy a completely risk-free portfolio (i.e., one that has the same payoff
in both states of nature), how many shares of j and k would you buy? (You may buy fractions
of shares.)
(b) What is the one-period risk-free rate of interest?
(c) If there were two securities and three states of nature, you would not be able to find a
completely risk-free portfolio. Why not?
6 Suppose that there are only two possible future states of the world, and the utility function is
logarithmic.19 Let the probability of state 1, π1, equal 23 , and the prices of the pure securities, p1
and p2, equal $0.60 and $0.40, respectively. An individual has an initial wealth or endowment,
W0, of $50,000.
(a) What amounts will the risk-averse individual invest in pure securities 1 and 2?
(b) How will the individual divide his or her initial endowment between current and future
consumption?
(Hint: Use the wealth constraint instead of the Lagrange multiplier technique.)

REFERENCES
Arrow, K. J., “The Role of Securities in the Optimal Allocation of Risk-Bearing,” Review of Economic Studies,
1964, 91–96.
———, Theory of Risk-Bearing. Markham, Chicago, 1971.

19 Problem 6 was suggested by Professor Herb Johnson of the University of California, Davis.

92
Edition
 State Preference Theory

Banz, R. W., and M. Miller, “Prices for State-Contingent Claims: Some Estimates and Applications,” Journal of
Business, October 1978, 653–672.
Breeden, D. T., and R. H. Litzenberger, “Prices of State-Contingent Claims Implicit in Option Prices,” Journal of
Business, October 1978, 621–651.
Brennan, M. J., and A. Kraus, “The Geometry of Separation and Myopia,” Journal of Financial and Quantitative
Analysis, June 1976, 171–193.
Cass, D., and J. E. Stiglitz, “The Structure of Investor Preferences and Asset Returns, and Separability in Portfolio
Allocation: A Contribution to the Pure Theory of Mutual Funds,” Journal of Economic Theory, June 1970,
122–160.
DeAngelo, H. C., “Competition and Unanimity,” American Economic Review, March 1981, 18–28.
DeAngelo, H. C., and R. W. Masulis, “Leverage and Dividend Irrelevance under Corporate and Personal Taxation,”
Journal of Finance, May 1980a, 453–464.
———, “Optimal Capital Structure under Corporate and Personal Taxation,” Journal of Financial Economics,
March 1980b, 3–29.
Debreu, G., The Theory of Value. Wiley, New York, 1959.
Dreze, J. H., “Market Allocation under Uncertainty,” European Economic Review, Winter 1970–1971, 133–165.
Fama, E. F., and M. H. Miller, The Theory of Finance. Holt, Rinehart, and Winston, New York, 1972.
Fisher, Irving, The Theory of Interest. Macmillan, London, 1930.
Garman, M., “The Pricing of Supershares,” Journal of Financial Economics, March 1978, 3–10.
Hirshleifer, J., “Efficient Allocation of Capital in an Uncertain World,” American Economic Review, May 1964,
77–85.
———, “Investment Decision under Uncertainty: Choice-Theoretic Approaches,” Quarterly Journal of Economics,
November 1965, 509–536.
———, “Investment Decision under Uncertainty: Application of the State-Preference Approach,” Quarterly
Journal of Economics, May 1966, 252–277.
—–, Investment, Interest, and Capital. Prentice-Hall, Englewood Cliffs, N.J., 1970, 215–276.
Kraus, A., and R. Litzenberger, “A State-Preference Model of Optimal Financial Leverage,” Journal of Finance,
September 1973, 911–922.
Krouse, C. G., Capital Markets and Prices. Elsevier Science Publishers B.V., Amsterdam, 1986.
Leland, H. E., “Production Theory and the Stock Market,” Bell Journal of Economics and Management Science,
Spring 1973, 141–183.
Merton, R., “The Theory of Rational Option Pricing,” Bell Journal of Economics and Management Science, 1974,
125–144.
Mossin, J., The Economic Efficiency of Financial Markets. D.C. Heath, Lexington, 1977, 21–40.
Myers, S. C., “A Time-State Preference Model of Security Valuation,” Journal of Financial and Quantitative
Analysis, March 1968, 1–33.
Ross, S. A., “Options and Efficiency,” Quarterly Journal of Economics, February 1976, 75–86.
———, “Return, Risk and Arbitrage,” in Friend and Bicksler, eds., Risk and Return in Finance, Volume 1. Ballinger
Publishing Company, Cambridge, Mass., 1977, 189–218.
———, “Mutual Fund Separation in Financial Theory—The Separating Distributions,” Journal of Economic
Theory, December 1978, 254–286.
Sharpe, W. F., Portfolio Theory and Capital Markets, Chapter 10, “State-Preference Theory.” McGraw-Hill, New
York, 1970, 202–222.
Stiglitz, J. E., “A Re-examination of the Modigliani-Miller Theorem,” American Economic Review, December
1969, 784–793.

93
Edition
 State Preference Theory

Appendix A. Forming a Portfolio of Pure Securities

If we have n market securities and n states of the world, and we are not sure the n market securities
are independent, we can find out by taking the determinant of the payoffs from the securities: a
nonzero determinant implies independence. For example, the set of pure securities is independent
since

1 0 0

0 1 0
= 1;

0 0 1

but

1 0 0

0 1 1
=
1 1
= 0



1 1

1 1 1

implies that the security payoffs (1, 0, 0), (0, 1, 1), and (1, 1, 1) are not linearly independent.
We can use Appendix: Matrix Algebra to form a portfolio of pure securities from an arbitrary
complete set of market securities. This involves computing the inverse of the payoff matrix for the
actual securities. For example, if (1, 0, 0), (0, 1, 1), and (0, 1, 3) are available, then define

⎛ ⎞
1 0 0
A=⎝0 1 1⎠
0 1 3

as the payoff matrix. Thus the determinant of A is

1 0 0

1 1

|A| =

0 1 1

= 2  = 0.

0 1 3
1 3

Let Xij be the amount of the j th security one buys in forming the ith pure security, and let X be
the matrix formed from the Xij . Then we require that
⎛ ⎞
1 0 0
XA = I where I =⎝0 1 0⎠
0 0 1

is the identity matrix and also the matrix of payoffs from the pure securities. Hence X = A−1. In
the present example
⎛ ⎞ ⎛ ⎞
2 0 0 1 0 0
1 3
− 21
A−1 = ⎝ 0 3 −1 ⎠ = ⎝ 0 2 ⎠.
2 0 −1 1 0 − 21 2
1

We then multiply X times A or equivalently A−1A to obtain a matrix of payoffs from the pure
securities. We have

94
Edition
 State Preference Theory

⎛ ⎞⎛ ⎞ ⎛ ⎞
1 0 0 1 0 0 1 0 0
⎝0 2
3
− 21 ⎠ ⎝ 0 1 1⎠ = ⎝ 0 1 0 ⎠ .
0 − 21 2
1 0 1 3 0 0 1

We can now see that the purpose of finding the inverse of A is to obtain directions for forming
a portfolio that will yield a matrix of payoffs from the pure securities—the identity matrix. Recall
that the three securities available are (1, 0, 0), (0, 1, 1), and (0, 1, 3). To obtain the pure security
payoff (1, 0, 0), we buy the security with that pattern of payoffs under the three states. To obtain
(0, 1, 0), we buy 23 of (0, 1, 1) and sell short 21 of (0, 1, 3). To obtain (0, 0, 1), we sell short 21 of
(0, 1, 1) and buy 21 of (0, 1, 3).

Appendix B. Use of Prices of State-Contingent Claims in

Capital Budgeting
Banz and Miller [1978] develop estimates of prices for state-contingent claims that can be applied
in capital budgeting problems. Similar methodologies were developed about the same time by
Garman [1978] and by Breeden and Litzenberger [1978].
Banz and Miller note that a fundamental breakthrough was provided in Ross [1976], who
demonstrated that by selling or buying options on a portfolio of existing securities, investors could
obtain any desired pattern of returns—“investors could span the return space to any degree of
fineness desired” [Banz and Miller, 1978, 658].
Banz and Miller present their estimates of state prices in a format similar to standard interest
tables. Like other interest tables, the estimates of state prices can in principle be used by any
firm in any industry or activity (subject to some important cautions and qualifications). Thus the
reciprocals (minus one) of the state prices computed are analogous to single-period interest rates.
Banz and Miller handle the multiperiod case by assuming stability in the relations between initial
states and outcome states. Thus the two-period matrix is simply the square of the one-period matrix
and the two-period matrix, the four-period matrix is the product of the one-period matrix and the
three-period matrix, and so on. In equation form,

V n = V · V n−1. (B.1)

The perpetuity matrix is the one-period matrix times the inverse of the identity matrix minus the
one-period matrix, or V (I − V )−1.
Their computations for a V matrix of real discount factors for three states of the world is provided
in Table B.1. In the definition of states in Table B.1 the state boundaries are defined over returns on
the market.
The conditional means are expected market returns under alternative states. The elements of
any matrix V may be interpreted by use of the first group of data. The .5251 represents the outcome
for state 1 when the initial state was also state 1. The .2935 represents the outcome for state 2 when
state 1 was the initial state. By analogy the .1612 represents an outcome for state 3 when state 3
was the initial state.
For equal probabilities the current price of a claim to funds in a state in which funds are scarce
(a depression) will be higher than in a boom state when returns are more favorable. Thus a project
with most of its payoffs contingent on a boom will have a lower value per dollar of expected returns
than a project whose payoffs are relatively more favorable during a depression.

95
Edition
 State Preference Theory

Table B.1 Three-State Matrix of State Prices and Matrix Powers

Definition of States
State State Boundaries ∗ Conditional Mean (R i,mr )
1 −.8647 − +.0006 −.1352
2 +.0006 − +.2042 +.0972
3 +.2042 − +1.7183† +.3854

State Prices
Implied Annual
Real Riskless
State 1 2 3 Row Sum Rate
1 year (V ):
1 .5251 .2935 .1735 .9921 .0079
2 .5398 .2912 .1672 .9982 .0018
3 .5544 .2888 .1612 1.0044 −.0044
2 years (V 2 ):
1 .5304 .2897 .1681 .9882 .0056
2 .5333 .2915 .1693 .9941 .0030
3 .5364 .2934 .1705 1.0003 −.0001
3 years (V 3):
1 .5281 .2886 .1676 .9843 .0053
2 .5313 .2903 .1686 .9902 .0033
3 .5345 .2921 .1696 .9962 .0013
4 years (V 4):
1 .5260 .2874 .1669 .9803 .0050
2 .5291 .2892 .1679 .9862 .0035
3 .5324 .2909 .1689 .9922 .0026
5 years (V 5):
1 .5239 .2863 .1662 .9764 .0048
2 .5270 .2880 .1672 .9822 .0036
3 .5302 .2897 .1682 .9881 .0024
6 years (V 6):
1 .5217 .2851 .1655 .9723 .0047
2 .5249 .2968 .1665 .9782 .0037
3 .5281 .2886 .1676 .9843 .0027
7 years (V 7):
1 .5197 .2840 .1649 .9685 .0046
2 .5228 .2857 .1659 .9744 .0043
3 .5260 .2874 .1669 .9803 .0033

96
 State Preference Theory

Table B.1 (continued)

8 years (V 8):
1 .5176 .2828 .1642 .9646 .0045
2 .5207 .2845 .1652 .9704 .0038
3 .5239 .2863 .1662 .9764 .0030
9 years (V 9):
1 .5155 .2817 .1636 .9608 .0045
2 .5186 .2823 .1639 .9627 .0038
3 .5197 .2840 .1649 .9686 .0031
10 years (V 10 ):
1 .5134 .2806 .1629 .9569 .0044
2 .5165 .2823 .1639 .9627 .0038
3 .5197 .2840 .1649 .9686 .0032
Perpetuity:
 
V (I − V )−1 :
1 132.50 72.41 42.05 246.96 .0040
2 133.31 72.85 42.30 248.46 .0040
3 134.14 73.29 42.55 249.98 .0040
* Chosen to yield ranges of Rmr that are approximately equally probable.
† Arbitrary truncations.

Table B.2 Cash Flow Patterns for an Investment

Range of Rates of Cash Flow before Steady-State
State of the Return on the Competition Cash Flow after
Economy Market Portfolio Enters X m Competition Enters Xc
Depression −.8647 − +.0006 300 −20
Normal +.0006 − +.2042 400 20
Boom +.2042 − +1.7183 500 40

The vector of gross present values of the project, Gk , will be

p

Gk = V t X k (t), (B.2)
t=1

where X k (t) is a vector whose elements represent the expected real cash flows of project k in
year t, assuming the economy is in state i. The summation is performed over time periods ending
in p, the last period during which the project’s cash flows are nonzero in any state.
An example of how the “interest factors” in Table B.1 can be applied is based on the illustration
presented by Banz and Miller. The Omega Corporation is analyzing an investment project whose
cash flow pattern in constant 1980 dollars (ignoring taxes and shields) is presented in Table B.2.

97
 State Preference Theory

The Banz-Miller example is sophisticated in illustrating that both the level and risk of the
cash flows vary with the degree of competition. In our example we modify their estimates of
the cumulative probabilities of competitive entry, using 0 in the year of introduction, .3 one year
later, .6 two years later, and 1 three years later. The risk-adjusted gross present value vector for the
project was set forth in Eq. (B.2). For the assumptions of our example, the particular gross present
value vector is
     
Gk = V X m + V 2 0.7X m + 0.3X c + V 3 0.4X m + 0.6X c + V 4 V (I − V )−1 X c .

We use the values of V and its powers as presented in Table 4B.1 to obtain the following results:
⎡ ⎤ ⎡ ⎤⎡ ⎤
gD .5251 .2935 .1735 300
⎣ gN ⎦ = ⎣ .5398 .2912 .1672 ⎦ ⎣ 400 ⎦
gB .5544 .2888 .1612 500
⎡ ⎤⎡ ⎤⎡ ⎤⎡ ⎤
.5304 .2897 .1681 204 .5281 .2886 .1676 108
+ ⎣ .5333 .2915 .1693 ⎦ ⎣ 286 ⎦ ⎣ .5313 .2903 .1686 ⎦ ⎣ 172 ⎦
.5364 .2934 .1705 362 .5345 .2921 .1696 224
⎡ ⎤⎡ ⎤⎡ ⎤
.5260 .2874 .1669 132.50 72.41 42.05 −20
+ ⎣ .5291 .2892 .1679 ⎦ ⎣ 133.31 72.85 42.30 ⎦ ⎣ 20 ⎦
.5324 .2909 .1689 134.14 73.29 42.55 40
⎡ ⎤
1230.09
= ⎣ 1235.68 ⎦
1241.48

If the initial investment were $1,236 in every state of the economy, the project would not have a
positive net present value if the economy were depressed or normal. However, the net present value
would be positive if the economy were strong. If initial investment costs had cyclical behavior,
particularly if supply bottlenecks developed during a boom, investment outlays might vary so
strongly with states of the world that net present values could be positive for a depressed economy
and negative for a booming economy.
The Banz-Miller use of state prices in capital budgeting is a promising application of the
state preference model. Further applications will provide additional tests of the feasibility of their
approach. More work in comparing the results under alternative approaches will provide increased
understanding of the advantages and possible limitations of the use of state prices as discount
factors in capital budgeting analysis.

98
 The results of a portfolio analysis
are no more than the logical
consequence of its information Objects of Choice:
concerning securities.

—Harry Markowitz, Portfolio Selection,

Mean-Variance
Yale University Press,
New Haven, 1959, 205. Portfolio Theory

Y OU ARE FAMILIAR WITH THE THEORY of how risk-averse investors make choices
in a world with uncertainty and how you can use a state preference framework to show
that the fundamental objects of choice are payoffs offered in different states of nature. While
this is a very general approach, it lacks empirical content. It would be difficult, if not impossible,
to list all payoffs offered in different states of nature. To provide a framework for analysis where
objects of choice are readily measurable, this chapter develops mean-variance objects of choice. Investors’
indifference curves are assumed to be defined in terms of the mean and variance of asset returns.
While much less general than state preference theory, the mean-variance portfolio theory introduced
here is statistical in nature and therefore lends itself to empirical testing.
One of the most important developments in finance theory in the last few decades is the ability
to talk about risk in a quantifiable fashion. If we know how to measure and price financial risk correctly,
we can properly value risky assets. This in turn leads to better allocation of resources in the economy.
Investors can do a better job of allocating their savings to various types of risky securities, and
managers can better allocate the funds provided by shareholders and creditors among scarce capital
resources.
This chapter begins with simple measures of risk and return for a single asset and then com-
plicates the discussion by moving to risk and return for a portfolio of many risky assets. Decision
rules are then developed to show how individuals choose optimal portfolios that maximize their
expected utility of wealth, first in a world without riskless borrowing and lending, then with such
opportunities.

From Chapter 5 of Financial Theory and Corporate Policy, Fourth Edition. Thomas E. Copeland, J. Fred Weston, Kuldeep

99