CAPM Basics

Volker B¨ ohm

Department of Economics University of Bielefeld P.O. Box 100 131 D–33501 Bielefeld, Germany e-mail: vboehm@wiwi.uni-bielefeld.de

December 2002
Discussion Paper No. 495

Abstract This paper discusses demand behavior of consumers and existence of equilibria for the standard capital asset pricing model (CAPM) with one riskless and finitely many risky assets, mean variance preferences of consumers, and subjective expectations. By treating individual expectations explicitly and parametrically, the model encompasses the description of individual as well as aggregate demand behavior for heterogeneous expectations. The paper provides a basic factorization formula for individual asset demand which implies the mutual fund property for agents with homogeneous expectations. This approach unveils some of the hidden structural features of the CAPM model often not discussed in the literature. Applying notions from standard static consumer theory, a characterization of the demand for risk from assumptions on risk preference is provided. The paper provides sufficient conditions on preferences to generate differentiable and globally invertible asset demand behavior of consumers parameterized by wealth and arbitrary subjective expectations. In addition, the paper proves existence, uniqueness, and determinacy of equilibria for the case of arbitrary homogeneous expectations, thus complementing, amending, and generalizing existing results. Examples indicate to what extent the conditions are necessary.

Key words: Capital asset pricing, mean variance preferences, asset demand, asset market equilibrium, existence, uniqueness, determinacy, expectations. JEL classification: E17, G12, O16.

Acknowledgement. This research is part of the project ‘Evolution auf Finanzm¨ arkten’ supported by the Deutsche Forschungsgemeinschaft under contract No. He 2592/2-1.

0

CONTENTS

1

Contents
1 Introduction 2 The 2.1 2.2 2.3 2.4 2.5 Basic CAPM Model Efficient Portfolios . . . . . . . . . . . . . . . . Asset demand with Mean–Variance Preferences Budget restrictions and short sale constraints . Existence of Demand . . . . . . . . . . . . . . . Differentiable Demand . . . . . . . . . . . . . . 2 3 5 7 9 10 15

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

3 Asset Market Equilibrium 19 3.1 Existence and Uniqueness of Equilibrium . . . . . . . . . . . . . . . . . . . 21 3.2 The equilibrium set and determinacy . . . . . . . . . . . . . . . . . . . . . 27 4 Equilibria with asset endowments 30 4.1 Homogeneous Beliefs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 5 Conclusion 33

1

1 Introduction

2

1

Introduction

The capital asset pricing model (CAPM)1 has become one of the most widely used models to describe demand behavior and equilibria in asset markets, deducing such well known results as the ’mutual fund theorem’ or the so called ’beta pricing’ relation. Although the terminology suggests that the pricing of assets is an integral part of the model, the results are mostly given on rates of returns rather than making asset prices fully explicit 2 .In addition, most presentations employ a static two period version of the model invoking the notion of a temporary asset market equilibrium within an environment of uncertainty for all agents, treating the equilibrium for given expectations which often are also not made explicit. In this case, the role of the expectations on demand and/or existence of temporary equilibria cannot be studied. Since the static model precludes a systematic analysis of the stochastic nature of the underlying random process on returns, the relationship, between the expectations, demand, and realizations cannot be studied as well. Few studies analyze the full stochastic embedding into an infinite sequential economy with exogenous random perturbations3 . In an attempt to embed the CAPM into a dynamic model with overlapping generations of agents (see B¨ ohm & Chiarella 2000), it was necessary to establish existence and uniqueness of temporary equilibria of the CAPM type model in an extended economic frame work with explicit parameterizations of incomes, prices, and expectations. In so doing, a multiplicative factorization of asset demand was obtained as well as some additional structural relationship of equilibrium asset prices were derived. Altogether these results shed some new light on the CAPM model and its usefulness for a dynamic analysis of asset markets. The model used here to describe the temporary situation of an economy is (at first) a simplified version of the standard two period model of an economy with given real wealth of agents, facing uncertainty about next periods realization of the real value of their portfolio consisting of real savings and of K assets decided on in the current period. The primary difference relative to the existing literature consists of making all parameters of the consumer decision problem i. e. prices and expectations explicit. By keeping them parametrically and conceptually separate in the formulation of the model the structural equations reveal their separate impact on individual asset demand and on equilibrium asset prices. Thus the functional relationship between expectations and equilibrium asset prices is unlocked revealing the different impact of expectations and of preferences between them. This provides an explicit formulation of efficient portfolios and of the mutual fund property for agents with homogeneous beliefs. At the same time, heterogeneous expectations as well as constraints on budget and short sales are identified as additional sources of inefficiencies. For the proof of existence and of uniqueness of equilibria this parametric separation also proves useful presenting a version of the beta pricing relation as a function of expectational parameters as well. While the method of proof here is different from most existing
as initially studied by Sharpe (1964), Lintner (1965), Mossin (1966) see LeRoy & Werner (2001) as one typical example. 3 Stapleton & Subrahmanyam (1978) is one of the most notable exceptions; for others see the discussion in B¨ ohm & Chiarella (2000)
2 1

2

Consumers/Investors Portfolio choices on the basis of mean–variance preferences constitute a widely accepted frame work to analyze decisions under risk and their associated markets. In the general equilibrium context this may yield explicit functional forms of the equilibrium price map. to analyze the long run behavior or prices. this approach has several advantages and special features. of learning. and other behavioral generalizations can be studied directly. and efficiency can be identified more directly than in the existing literature. Lintner (1965). and returns. the conditions and assumptions here have their counterparts in those of the literature. the explicit parametric treatment allows the description of features of the equilibrium manifold. the beliefs on premia (q a − Rp. e.2 The Basic CAPM Model 3 ones4 . 1990. which make a full dynamic analysis possible. In other words. to study. Market interaction occurs among finitely many investors who do not possess storage possibilities for the consumption good directly and who attempt to maximize the expected utility of future consumption (wealth). and Mossin (1966) 4 3 . V a ) determine the mix of risky assets (V a )−1 π a of any optimal portfolio. a consumption commodity which serves as numeraire and as a riskless asset with interest factor R > 0. i. and exhibit the properties of rational expectations equilibria. In other words. prove. demand ϕa (λπ a ) and the vector of subjective conditional premia (V a )−1 π a are linearly dependent. portfolios. their role for existence. Since these are made on the primitives of the agents’ characteristics. In other more general situations. the safe rate R and initial wealth Re a . Moreover. 1989. while the level of asset demand h(π a ) = ma (ρa )/ρa is determined by the utility function. As a consequence. Nielsen (1988). They form the basis of the classical capital asset pricing model (CAPM) the results of which serve as a fundamental guideline to understanding and evaluating the trade off between returns and risk in asset markets The primary importance of mean–variance preferences within the classical asset pricing theory stems from the fact that they supply a convenient structure to analyze asset demand behavior explicitly. 1987. mean–variance preferences induce globally invertible demand functions which are often solvable algebraically. inducing a much studied class of models successfully used in financial theory. which are needed if an explicit description of the asset price process is Without claiming to be complete. as is known with quadratic utility. uniqueness. some of the most noted are. Werner (1985). 2 The Basic CAPM Model The model describes markets for K ≥ 1 risky assets. The case with quadratic utility and normally distributed returns yields the well known standard CAPM pricing formula of Sharpe–Lintner–Mossin 5 . Laitenberger & L¨ offler (2002) 5 Sharpe (1964). for all π a and all λ ∈ R+ . and Hens. . In addition the impact of different and heterogeneous expectations rules. 1988. Allingham (1991) Dana (1999). Together with concavity of the mean–variance utility function one finds that individual asset demand ϕa is globally invertible (see Lemma ?? in the appendix).

·)] ν (dq ) 2 x. p. x + y = k=1 p(k) x(k) + y. ·)] = with associated subjective variance Vν [W (x. Concerning subjective expectations6 . for any asset portfolio x ∈ RK this yields the subjectively expected value of his future wealth Eν [W (x. characterizing the expectations subjectively held by the investor. q ) − Eν [W (x. V x . w. w. 6 see Nielsen (1988) 4 . x(K ) ). Then. y ) ∈ RK × R denote a portfolio of K + 1 assets and let p ∈ RK denote current asset prices in units of the numeraire commodity. w. σ ) →R → U (µ.. which implies that the consumer assumes that non of the assets is redundant. the following assumption will be made throughout. σ ) which is increasing in the mean µ and decreasing in the standard deviation σ .2 The Basic CAPM Model 4 the goal. p. q ) = R[w − p. x = Rw + (q − Rp). . V ) ∈ RK × RK of expected mean 2 cum dividend prices q e ∈ RK and associated covariance matrix V ∈ RK . x Rw + (q − Rp). Let ν ∈ P denote a probability measure on RK describing the distribution of future cum dividend prices. x ] + q. In this context. . . x W (x. without (short sale) constraints on the demand for all K + 1 assets the investors budget constraint requires K w = p. Future Wealth Let (x(1) . . p. x ν (dq ) = Rw + (q e − Rp). denotes the scalar product of any two vectors. . Then. w. ·)] = = RK + RK where . p. w. y = (x. • a subjective probability distribution ν ∈ Prob(RK ) for future cum-dividend asset 2 prices (gross returns) parameterized by a pair (q e . If q e := p + d ∈ RK denotes future cum-dividend prices of the K assets his future (random) consumption/wealth becomes W (x. the typical consumer/investor is characterized by • w ∈ R units of a numeraire commodity as his consumption/numeraire endowment • a one period planning horizon • risk preferences over the mean µ and the standard deviation σ of his future consumption/wealth described by a utility function U: R × R+ (µ. p.

x . one often writes Xef f instead of Xef f (π. p. since it depends on individual subjective beliefs. V ).2. V ). x. V x x∈RK 1 2 for every (w. V ) are given. V x ≥ x V x with at least one strict inequality. When asset prices p and expectations (q e . As a consequence asset demand of an investor with µ − σ −preferences and expectations ν can be defined as ϕ(w. Vν [W (x. ·)] 2 x∈RK 1 2 (2. typically. x ≥ π.1 Xef f (π. π.1) (2. Note that this concept of efficiency is strictly subjective and agent specific.5) (2.2) = arg max U Rw0 + π.1 2. Lemma 2. (ii) V ∈ RK is symmetric and positive definite. V x 2 . ·)].1 An asset portfolio x ∈ RK is called mean–variance–efficient (or µ − σ –efficient) with respect to (π. V ) denote the set of all µ − σ –efficient portfolios. one simply speaks of efficiency rather than of µ − σ efficiency.1 Efficient Portfolios 5 Assumption 2. x .1 Efficient Portfolios Definition 2. V ) satisfying assumption 2. ·)]. p. if there is no x = x such that π. Define π := q e − Rp as the vector of expected risk premia of the K assets. Similarly. π. p. ·)] 2 = U Rw + π. Vν [W (x. Then. w. a particular portfolio x will not be efficient for agents with heterogeneous beliefs. π = 0. x. implying that. w. Let Xef f (π.1 The subjective expectations for future cum dividend prices are parameterized in • q e = pe + de ∈ RK the vector of expected values of future cum dividend prices and • V ∈ RK ×K the associated covariance matrix and satisfy: (i) q ∈ RK . x x.3) (2.4) . V ) = λV −1 π |λ ≥ 0 5 (2. p. w. V ) := arg max U Eν [W (x. w. one can write 1 1 U Eν [W (x.

V x − λ V −1 π. V λV −1 π < x. which implies λV −1 π.2. V λV −1 π < x. x = π. λV −1 π . x x. It suffices to show that λV −1 π. V x − λV −1 π. V x x.1 Efficient Portfolios 6 Proof: Consider λ > 0 and some x = λV −1 π with π. V x − λ π. V x One has 0 < = = = = (x − λV −1 π ). V (x − λV −1 π ) x. V x ¦ ©   ¢¡¤£¥£  § ¦¨§ ¦. λV −1 π x. V λV −1 π QED. V x − λ π.

Thus. but they differ in their scale. π.6) . x . V x with the contour lines of the expected return π. Larger asset bundles imply larger means and larger variances. all efficient portfolios have the same mix of assets. It consists of all tangency points of the contours of the quadratic mapping x. x | x. V ) := max x∈RK π.1: Efficient Portfolios Figure (2. V x 6 1 2 ≤ σ = π.1) displays the set of efficient portfolios Xef f for the case of two assets. ¦  §  Figure 2. since they are positively homogeneous functions of the scale factor λ. V −1 π 2 σ 1 (2. It is immediate that the set of efficient portfolios can also be obtained as the family of solutions of the following standard optimization problem: m(σ.

V −1 π 2 is the marginal or shadow return to risk. i. If V ∈ RK ×K is positive definite. This means that the mix of asset demand is determined by prices and expectations while the level of demand is determined by preferences and wealth. x∈RK 1 (2. any demand vector is colinear to the efficient vector V −1 π .13) = λ π.2 Asset demand with Mean–Variance Preferences 7 Solutions to (2. Proposition 2. V −1 π = λ2 π. x . σ ≥ 0 (2. e. V x 2 ) ⊂ Xef f (π. V x 1 2 (2. all wealth and preference 7 .2 Asset demand with Mean–Variance Preferences Assumption 2. then ϕ(w. such that : x = λV −1 π π. x x. The next two propositions state some fundamental properties of asset demand. V −1 π 1 2 In other words.6) exist whenever V is positive definite. V −1 π 1 2 ≤σ = σ π. V −1 π = π. V x π.2 Risk preferences are described by a continuous quasi concave utility function of mean and standard deviation (µ. Its graph is the so called 1 efficiency line and its slope π.12) (2. π.2) be satisfied. V ). σ ) R × R+ → R U: (µ.1 Let assumption (2. x x. V ) := arg max u(Rw + π. x | x.8) For given (π. They are unique if π = 0 and given by ψ (σ. the function (??) defines a linear relationship between expected mean µ and the standard deviation σ among efficient portfolios.9) For every x ∈ ϕ(w. non decreasing in its first argument µ and non increasing in its second argument σ .7) which yields Xef f = x |x = 1 2 V −1 π. In some situations it may be convenient to include preferences which are monotonic only in a weak sense. 2. V −1 π 1 2 V −1 π (2. there exists λ ≥ 0.11) (2. σ ) which is strictly increasing in its first argument µ and strictly decreasing in its second argument σ . σ ) → U (µ. π. V ). V ).2. x. π. V ) := arg max x∈RK π. V x σ π.10) (2. Thus.

V ). V ) ϕ(w.15) (2. V ). π. V −1 π = λ2 π. V ϕ(w. V x 2 )   © £¥¤§¦¨¦   .10) QED.2) provides a geometric characterization of an efficient demand bundle x∗ for the situation with two assets. π. V ) ϕ(w.10) follows immediately from the efficiency lemma (2. V −1 π = π. ϕ(w. ϕ(w. π.13) imply that the demand function ϕ(w. V ϕ(w. V ) 1 2 (2. π. All other properties are direct implications of (2. π. V ) : ϕ(w. Proof: Property (2. π. V ) satisfies the following properties hold for all (w.16) (2. π. π. π.2 Asset demand with Mean–Variance Preferences 8 characteristics must be embodied in the factor λ of equation (2. x. the properties (2. V ) π.10) as a function of (w. V ). x .10).17) = λ π. V −1 π Figure (2.(2. If demand is unique. V ) = λV −1 π π.14) (2.1) have been superimposed 1 by the contours of the function of feasible utilities U (Rw0 + π. π.2. The iso–variance lines of figure (2.1).

Combining these findings one can now derive the 8 . Moreover.  ¢¡ Figure 2. it is also the parametrically given (linear) shadow price of the efficiency solution (2.6). This is equivalent to the statement of equation (2. V −1 π . It is simultaneously equal to the mean and to the variance of the efficient portfolio V −1 π .13) that efficient µ − σ combinations have a constant ratio of π. π. V −1 π plays an important role in these equations. V −1 π is the marginal (shadow) return to any additional risk given prices and expectations.2: Efficient demand Notice that the number π. which is a homogeneous function of degree two in prices and expectations. In other words.

when efficiency to buy short. 2. V −1 π 2 ) π. V −1 π 1 2 1 V −1 π. V −1 π 2 ) := arg max U (Rw + π. e. V −1 π 2 ). V ) efficient. π. V −1 π 2 σ. π. V −1 π 2 ). the individual level of asset demand is equal to the ratio of the indirect standard deviation and that of the standardized portfolio V −1 π .18) s(w. creating a wedge between the efficiency trade off and the risk trade off. V x 1 2 = x∈Xef f σ ≥0 max U Rw + π. This result also confirms the so called mutual fund theorem. s(w. σ ). It is well known that they will cause spillovers in demand between different assets. Proof: One verifies easily that x∈RK max U Rw + π. σ ≥0 1 1 (2.19) Therefore. π. π.3 Budget restrictions and short sale constraints 9 precise expression of the scale of asset demand λ to obtain the fundamental factorization formula of asset demand. V ) = s(w. demand an agent deviates from the efficiency line depending on individual preferences. V x 1 2 = max U (Rw + m(σ. π. In such a case the constraints act like quantity rationing constraints. As a consequence.2) suggests. e. V −1 π 2 σ. π. Thus. π. V ) := arg max U w + π. x. i. V −1 π 2 ) π. demand will not be (π. aggregate demand in any market situation with homogeneous beliefs will not be efficient if some agent faces a binding short sales constraint. It is evident that such situations are also not constraint efficient in general7 . agents with identical beliefs hold identical mixes of assets. QED. x .2. σ ) σ ≥0 1 1 1 Therefore: ϕ(w.2 ϕ(w. V ) := ψ (s(w. x . π. σ ) = U (Rw + m(s(w. π. Lemma 2. V −1 π 1 2 1 1 V −1 π. as the geometry of Figure (2.3 Budget restrictions and short sale constraints Short sales constraints may apply when Xef f ∈ / RK + . In other words. Binding short sales constraints correspond exactly to quantity constraints as treated in the rationing literature. x. V −1 π 2 ) = max U (Rw + π. (2. x . x. V x x∈RK 1 2 = where s(w. demand with a binding short sale constraint will no longer be colinear to V −1 π in general. V ). i. Therefore all results concerning efficiency and constraint efficiency carry over directly to the asset demand here (see for example B¨ ohm & M¨ uller 1977) 7 9 . π. Therefore.

 ¢¡¤£¥£  § §©¨    § PSfrag replacements   §. but p. x ˜∈ / Xef f ) occurs when prices p and expectation q are not colinear.4 Existence of Demand 10 A similar loss of efficiency occurs when the budget constraint of an agent becomes binding. e. i. Figure ?? describes such a situation with two assets without short sales constraint. e. The inefficiency (i. x ∗ > w. when he is not allowed to go short on the safe asset to obtain credit.2.

V ) = 1 −1 1 V π = V −1 (q e − Rp) α α (2. π. V −1 π ) α 10 .20) 2 where α is usually interpreted as a measure of risk aversion. ϕ(w.4 Existence of Demand The following four examples provide additional insight into possible properties of asset demand.21) Thus.3: Inefficiency of Demand x ˜ 2. From the first order conditions one calculates directly the demand as U (µ. V ) := π. π. σ ) = µ − ϕ(w. π. R(1) R(0) g replacements R(1) R(0) ¦   (a) Inefficient demand x ˜ with short sale constraint (b) Inefficient demand x ˜ with binding budget constraint Figure 2. and independent of initial wealth. They also indicate which additional assumptions are needed to obtain well defined demand for all prices and expectations. V ) = ( π.22) m(w. Notice that the investor’s (indirect) expected return is from his asset demand 1 (2. σ ) →R → U (µ. σ ) α 2 σ . (2. asset demand is linear (homogeneous of degree one) and globally invertible in π . Example 1: Linear mean variance preferences Let the preferences of an investor U: be given by the function R × R+ (µ.

As a consequence. π.24) (αRw)2 + 4α π. π. π. the demand for risk is decreasing in wealth w. prices.28) √ For w = 0 the investor will always demand constant risk equal to 1 / α. V −1 π 4( π. The homogeneity of demand implies that the investor would be willing to bear unlimited risk when the ratio of expected return to risk of the standardized portfolio becomes unbounded.29) . π.27) = 1 β2 (2. α (2. his asset demand will be bounded uniformly for all (w. ϕ(w. V −1 π β One obtains as the investors expected return m(w. V ) 1 1 2 = 1 1 π. V −1 π 2 ) := ϕ(w. Moreover. V −1 π ) (2. r + σβ r > 0. π. V ). V ϕ(w. Example 2: Logarithmic mean utility Consider next the situation with the utility function U (µ. 1 π. which implies that his demand for risk is √ always less than or equal to α. Example 3: Quasi concave utility Consider next the situation with the utility function U (µ. V ) = and his demand for risk as s(w.23) implying that he always chooses a standard deviation equal to his risk factor 1/α (some times called the risk tolerance) times the risk of the standardized portfolio which corresponds to efficient shadow return to risk. σ ) = ln(µ) − which yields the asset demand function ϕ(w. π. β (2. π. regardless of wealth. V )). for √ w > 0. σ ) = µα . one observes that β ≥ α for all π . π. and expectations. V ) := π. V ).2.25) α 2 σ 2 µ > 0. (2. Moreover. V −1 π − αβRw = 0. V ) = where β= (αRw + 1 −1 V π. V ) := ϕ(w. V ) 1 2 1 2 (2. 11 0<α≤1≤β µ≥0 (2. V ϕ(w. π.26) β is the unique positive solution of the quadratic equation (β 2 − α ) π. V −1 π 2 .4 Existence of Demand 11 and his demand for risk (his indirect standard deviation) is s( π.

V ) = Xef f . U (λ) is strictly increasing in λ for π. V −1 π 1 2 λ2 π. consider an arbitrary λ > 0 and an associated efficient bundle x(λ) = λV −1 π . 1 Example 4: Unbounded Demand Finally. π.   r   Rw 1 (2. if π. For π. V ) = 1 π. V −1 π 2 . V −1 π π. V −1 π 2 < 1. but strictly quasi concave as long as α < β . V −1 π 1 2 1 2 1 2 (2.31) ϕ(R. If α = β = 1. Therefore. asset demand does 1 2 1 − π. It will be shown that the demand for risk and the demand are not defined for a large range of prices and expectations. π. This induces a feasible utility U (λ) := Rw + λ π.2. V −1 π 2 < r Thus. independent of w and π. e. V −1 π 2 is larger than Rw/r.32) which is also strictly quasi concave.34) −1 . if π. V −1 π π. i. V ). If it exists it is either zero or the whole set of efficient asset bundles Xef f . r+σ r > 0. Thus. One finds  Rw 1  −1  2 > ∅ . Therefore.  r    Rw 1  0. one obtains r 1+ ϕ(w. V −1 π 1 not exist. V ) for α < β . V ) = ≥ 1.33) (2. V −1 π )= 1 1 2 rα β−α 1 β is constant. if π. V −1 π − > Rw + λ π. V −1 π 2 = . V 12 −1 π 1 2 −1 V −1 π (2. V π . π.35) .4 Existence of Demand 12 which is not concave. the indifference curves of U are rotating lines with higher slopes for higher utility. For w = 0. consider the situation with the concave utility function U (µ. V −1 π 1 2 rα β−α 1 β V −1 π (2. π. . (2. Given (π. demand is empty if the return to risk π.30) Notice again that the investor’s demand for risk s(0. σ ) = µ − σ2 . V −1 π r + λ π. asset demand is uniformly bounded for all (π. this yields the asset demand function ϕ(0.

σ | µ ≤ π. 1) | λ ≥ 0} 1 (2. s(w. V −1 π 2 )) 1 must lie in P (Rw. σ ≥ 0). σ ) | µ ≤ π. Definition 2. which is the intersection of two closed convex sets. 0) = {λ (0. V −1 π 2 σ. In other words. π. since asset demand becomes unbounded if and only if the demand for risk becomes unbounded for some finite shadow return for risk. σ ¯ ) := {(µσ )|U (µ. V −1 π 2 σ. Since. V −1 π 2 σ.2) show that properties of the utility function/the preferences and not expectations are responsible for the failure of demand to exist. e. V −1 π 2 ) := arg max u(Rw0 + µ.2 Let S denote a subset of Rm . 1) | λ ≥ 0} ∩ (µ. V −1 π 2 ). π. σ ) = ( π. V −1 π 2 s(w. σ ≥0 1 1 1 1 (2. V −1 π 2 ) is finite for all prices and expectations satisfying assumption (2. denoted AS is defined as ∞ AS := k=1 C (S k ) (2. σ ) ≥ U (¯ µ. It is compact if the intersection of their asymptotic cones is {0} 9 .40) Note the relationship to the discussion of useful trade vectors by Werner (1987) for details see Debreu (1959) or Hildenbrand (1974) 13 . Proof: 1 Monotonicity of the utility function allows one to rewrite the demand for risk s(Rw 0 . 1) alone. i. σ ¯ )} as the upper contour set for the utility function U (µ.36) where C (S k ) denotes the smallest closed cone containing S k := {x ∈ S | |x| ≥ k }. σ ¯ ).38) 1 Observe that any demand pair (µ.2. demand is nonempty .4 Existence of Demand 13 One observes that the demand for risk and therefore asset demand becomes unbounded 1 as π. V −1 π 2 approaches 1 from below. π. The asymptotic cone of S . If AP (Rw. V −1 π 2 ) as the solution of s(w. π. the set of convex preferences has to be restricted if well defined demand for all prices and expectations is required. σ ) | µ ≤ π. if the asymptotic cone of the better set contains no strictly positive vectors 8 of R2 + . AP (Rw.2 be satisfied and define P (¯ µ. σ ≥ 0 = {0} 8 9 1 1 (2.37) then the demand for risk s(w. σ ≥ 0 . 1) | λ ≥ 0} it follows that AP (Rw0 .3 Let assumption 2. σ ) | µ ≤ π. π. σ ) at the point of (¯ µ.1). it consists of the half line through (0.39) (2. σ ≥ 0 = {λ(0. 0) ∩ A (µ. 0) = {λ (0. The last two examples combined with the factorization lemma (2. V −1 π 2 σ. Thus. Lemma 2. This leads in a natural way to the next lemma. 0) ∩ (µ.

More specifically. (Rw.1). π. uniqueness. and (2. 0). Moreover. and positivity of demand under mean-variance preferences. in other words a condition is needed which induces 1 non zero demand for risk for all positive shadow returns π. π. V −1 π 2 ) π. in addition. (2. V −1 π > 0. V −1 π 2 > 0. It follows from results of standard consumer theory. π. and continuity of demand. one obtains the following theorem on existence. it is evident that demand for risk will be positive 1 for all π. V ) = s(w.1) and assume that preferences are strictly quasi concave satisfying assumption (2. h(Rw. However. 0)} . σ →∞ σ Exploiting the curvature of h near zero. which implies a non zero asset demand. σ ) − Rw = 0. σ )) = U (Rw. Since U is strictly increasing in µ. the Inada conditions (a) h(Rw. Positive demand for risk follows from part (a) of the Inada condition. Theorem 2.2). σ ) − Rw lim = ∞. σ ) | U ((µ. σ →0 σ lim These two properties are a direct translation of the so called weak Inada condition used in growth models to convex functions. Therefore. then. If.3). σ ) − Rw =0 σ →0 σ lim 1 2 and (b) h(Rw. continuity of the utility function U implies the existence of demand. 0). and convex in σ . there exists a function h : R × R+ → R. 1)} ⊂ R 2 if and only if h(Rw. σ ). QED.4 Existence of Demand 14 QED.2). describing the indifference curve through (Rw. 14 . 0) = Rw0 for all Rw. Consider the utility contour associated with the initial wealth Rw {(µ. σ ) → h(Rw. one additional property is required to obtain non zero demand. Combining these observations with the results from lemmas (2. σ ) − Rw = ∞.41) together with strict quasi concavity implies existence. V −1 π 1 2 1 1 V −1 π = 0 Proof: Part (b) of the Inada conditions (2.2. V −1 π 2 ) > 0 (iii) asset demand ϕ(w. (i) asset demand is a continuous function for all π = 0 (ii) the demand for risk s(w. V −1 π 2 > 0 if and only if h(Rw. AP (Rw. 0) is equal to the half line {λ (0. h is continuous. for all π. uniqueness. σ →∞ σ lim (2. monotonically increasing. It is evident that the curvature of the function h is related to the properties of the asymptotic cone of P (Rw.1 Let expectations satisfy (2. that strict convexity of preferences implies that demand is a continuous function.41) hold.

we will continue to speak of demand for risk rather than supply. for example B¨ ohm & Barten (1982). Katzner (1968)). (see Katzner (1968) and Debreu (1982)). Assumption 2. Due to the factorization lemma (2. (w. The demand for risk induces a demand for expected return as m(w. σ ) | µ ≤ ρσ + Rw} .5 Differentiable Demand It is now straightforward to derive conditions under which one obtains differentiable demand functions by applying results from standard consumer theory. V −1 π 2 > 0 as the shadow return (or equivalently as the standard deviation of the standardized portfolio V −1 π .2) it is sufficient to derive a differentiable demand function for risk. σ ) →R → U (µ. which in turn implies the differentiability of the demand function for risk s : R × R + → R+ . σ ) be strictly quasi concave. For 1 given expectations and prices define ρ := π.43) (2.2) as well as U11 U12 U1 U21 U22 −U2 U 1 −U 2 0 for all interior (µ. ρ). ρ) := arg max {U (µ.5 Differentiable Demand 15 2. Since the utility is decreasing in risk equation (2. twice continuously differentiable and satisfy (2. Lemma 2.2.44) 15 .42) corresponds to a consumer’s supply of a factor to the market (like labor with real wage ρ). To obtain a differentiable demand function for risk the following additional assumption is necessary and sufficient (cif. σ ).44) of a non vanishing bordered Hessian matrix implies that the Gaussian curvature of all indifference curves is different from zero at all interior points. all further results on individual demand for risk have their complete analogues in consumer theory with endogenous labor supply allowing a direct application of all known results. ρ) → s(w. The condition (2.3 Let the utility function U: R × R+ (µ. As a consequence. Debreu (1972). ρ) (2. ρ) := Rw + ρs(w. where ρ is the price/return for risk taking. Then. the follow=0 (2. the demand for risk is given by s(w.4 Given the assumptions to guarantee differentiable demand for risk and return. Nevertheless.42) This maximization corresponds to a standard consumer problem with a linear budget set.

which corresponds to the micro economic analogue of a household’s labor supply. e. ρ) ∂w ∂w ∂m ρ ∂s (w. ρ) (w. ρ) s(w. The Slutzky equation (2. ρ) ∂ρ ∂s ∂m (w. which is bounded below. ∂s/∂w > 0. ∂s/∂ρ > 0. ρ) ∂ρ ∂w (2. risk normality may not be the most widely accepted economic assumption to be used for the CAPM model. By analogy from standard micro economic terminology. a property which is necessary for unique equilibria (see Section ?? below). Equation (2.47) Proof: The first two equations are identities induced by monotonicity and the budget equation. If wealth has a positive effect on the chosen amount of risk. ρ) > s(w.ρ (w.47) and risk normality imply that a higher shadow return to risk implies a higher chosen amount of risk.45) (2. It is difficult to obtain intuitive economic assumptions for general mean variance preferences which imply monotonicity of demand for risk in general. Notice.2. Notice that what is interpreted here as demand for risk is essentially the supply of a willingness to accept risk. define risk as an inferior good if ∂s/∂w < 011 . ρ) = s(w. ρ): ∂m ∂s (w. ρ) 1 + (w. ρ) ∂ρ s(w. it seems desirable to elucidate those assumption under risk inferiority for which demand for risk is still globally increasing. Combining (2. ρ) := ρ ∂s (w. QED. However. Moreover. ρ) − R > −R ∂w ∂w (2. Laitenberger & L¨ offler (2002) 16 . that with this terminology. this fact expresses a higher desire for risk with higher wealth. for separable utility functions one obtains the following result. ρ) = (w. allowing for positive as well as negative price effects. Therefore.ρ (w. Therefore. 10 11 Hens.45) – (2.47) one obtains Eσ. risk normality corresponds to the situation of labor supply increasing in wealth implying that leisure is an inferior good (a situation usually not considered as typical – or normal !). ρ) > ρ if ∂m/∂w > 0 where Eσ.5 Differentiable Demand 16 ing properties hold for all interior (w. demand for risk is globally increasing in the shadow return. Clearly.47) follows from standard duality and the Slutzky equation. risk inferiority is only a necessary condition for a decreasing demand for risk in the shadow return ρ.46) (2. Laitenberger & L¨ offler (2002) call this property decreasing (non increasing) risk aversion. ρ) ∂ρ ∂s ∂s (w. i. called increasing risk aversion by Hens. Using traditional terminology in this situation. ρ) = R + ρ (w. define risk to be a normal good10 if ∂s/∂w ≥ 0.48) defines the elasticity of the demand for risk with respect to ρ.

ρ)) + ρs(w.5 Differentiable Demand 17 Theorem 2. (v) i.54) i. (iii) ρ→∞ (2. whenever u is strictly concave. Standard economic 17 . ∂w v (s(w.52) (2. σ ) = u(µ) − v (σ ) satisfying (2. In such cases. risk is an inferior good. ρ) = . Regarding wealth effects they imply that the associated expansion paths are never up ward sloping. the demand for risk is increasing in the shadow return ρ.2. ρ)) − ρ2 u (Rw + ρs(w. but they represent the subtle balance between wealth effects and substitution effects to induce monotonic demand. they integrate the quasi linear utility and the strict concavity into one uniform set of assumptions.55) These relatively strong requirements for separable µ − σ preferences are somewhat surprising.51). ρ)) − ρ2 u (Rw + ρs(w. ρ) = 0 if and only if ρ = 0. e. e.50) strictly increasing in ∆ ≥ 0 and ∆→∞ lim ∆ u (µ + ∆) = ∞ (2. (ii) s(w. and v (σ ) > 0 for all σ > 0. ρ)) (2. since these violate the assumptions (2. (i) the demand for risk s(w. (2.49) lim v (σ ) = ∞.2) and v (0) = 0. u (Rw + ρs(w. ρ)) ∂ s(w. the demand for risk will not be monotonic in the shadow return (see Figure 3. ρ) = ∞ for every w ∈ R (iv) ∂ Ru (Rw + ρs(w. ρ)) s(w.5). If for all µ ∈ R: ∆ (ua ) (µ + ∆) hold. It is also somewhat unexpected that the sign of the price effect on the demand for risk is exclusively determined by second order properties of the function u. ρ) := arg max u(Rw + ρσ ) − v (σ ) σ σ →∞ (2. As the proof below shows.51) is well defined and continuously differentiable in (w. ρ)) (2. At the same time.53) lim s(w.2 Consider a separable concave twice continuously differentiable utility function U : R × R+ → R of the form U (µ. Observe that the theorem does not cover the case with piece wise quasi linear functions u. ρ)u (Rw + ρs(w. The two situations of quasi linearity (u = 0) and (v = 0) are boundary cases. they are essentially necessary. then. ρ). ρ) = >0 ∂ρ v (s(w.

the demand for risk s(w. rho)) = u (Rw + ρs(w.53) follows from the first order conditions 0= ∂H v (s(w. σ ) = 0 if and only if σ = 0. QED. ρ) is a function. If for all µ ≥ 0 µu (µ) is strictly increasing (2. Proof: The function H (ρ. while monotonicity of demand (2. The assumption (2. then ∂H (ρ. If ρ > 0.5 Differentiable Demand 18 intuition would suggest that the function v also captures part of an agent’s attitude toward risk. ∂σ since v (0) = 0 which proves (2. s(w.56) and µ→∞ lim µ u (µ) = ∞ (2. Inferiority of risk (2. Since ∂H (ρ. Property (2.1 Consider the separable case of theorem (??) and assume conditions (2. Corollary 2.2) and (3.55) follows from condition (2.51) implies that u is convex but that its steepness is bounded. s(w. ρ)) . ·) : R + → R+ is monotonically increasing and bijective in ρ on R+ . σ ) := u(Rw + ρσ ) − v (σ ) is strictly concave in σ for every ρ ≥ 0. Whether this is true in general still is an open question. Laitenberger & L¨ offler (2002)). which seems counterintuitive to the result that v has no influence on the sign of the derivative of demand for risk12 . s(w.57) hold. ρ)) ρ − ∂σ u (Rw + ρs(w.52). If this is case one has the following corollary. Hence.14).2. ρ)) (ρ.51). This confirms essentially for the separable case that measures of risk aversion which use properties of the function u alone are sufficient in capturing all relevant effects. one finds and from (2. (compare Hens. The strict convexity of the function v implies assumption (2. then. ρ) is differentiable. Notice that no restriction for the wealth to be positive is required. σ ) = 0 ∂σ has exactly one positive solution.54) follows directly from strict concavity and an application of the implicit function theorem.51).44). Hence. ∂σ ∂H (0. 12 18 . σ ) = ρu (Rw + ρσ ) − v (σ ). for given non negative wealth w ≥ 0.

53). the condition on the strong marginal disutility of risk as in (2.60 on the elasticity of the derivative of u was used also by Dana (1999). ρ)) − ρ2 u (Rw + ρs(w.3 Asset Market Equilibrium 19 Proof: Equation (2. ρ))u (Rw + ρs(w. r > 0 implies (2.56) but not (2. ρ)) Eu (µ) := µu (µ) > −1 for all µ ≥ 0.57) 19 . Note. implying the usual star shaped form and allowing for inferiority of both goods as well as Giffen effects. Figures 3. ρ)) − ρ u (Rw + ρs(w. However. ρ)) − ρ u (Rw + ρs(w. In summary. Let 0 = x ¯ ∈ RK denote aggregate asset supply. ρ)) 1 + Eu (Rw + ρs(w.2.2). ρ)) (Rw + ρs(w.55) implies ∂ u (Rw + ρs(w. and beliefs (q a .58) u (Rw + ρs(w. The method of proof also reveals that a joint condition like (2.51) is needed to generate unbounded demand for risk.4 display three possible offer curves violating theorem (2. ρ) = 1+ 2 ∂ρ v (s(w. However. their curvature is restricted only by quasi concavity.60) QED.51) is required to guarantee surjectivity of the demand for risk (2. ρ)) u (Rw + ρs(w. Thus.56) is equivalent to (2. ρ)) v (s(w. ρ)) ρs(w. This includes the quasi linear case of a (weakly) concave utility to expected return. since then risk becomes an inferior good. and thus whether the associated offer curve is monotonic. It is not sufficient to assume that the marginal utility of return tends to zero with infinite shadow return to risk. The demand for risk is always monotonically increasing in the shadow return and tending to infinity if risk is a normal good even for non separable preferences. that an offer curve can be backward bending for a large class of concave utility function. Notice. the curvature of u alone determines the sign of the price effect. 3. since the linear case is not a direct boundary situation of the strictly concave situation. ρ)) u (Rw + ρs(w. initial wealth w a . V a ). and 3.3. The assumption 2. ρ)) 1+ 2 v (s(w. ρ)) u (Rw + ρs(w. u ( µ) (2. ρ)) ≥ = (2. the two situations of a linear and of a strictly concave utility function u seem to require a different treatment when surjectivity is required. ρ)u (Rw + ρs(w.59) since the assumption (2. since it cannot be excluded that the demand for risk remains bounded. 3 Asset Market Equilibrium Let each consumers a ∈ A now be characterized by mean–variance preferences U a . that the elasticity assumption alone does not guarantee the surjectivity of risk demand13 . if the utility of expected return is not linear (strictly concave or piece wise linear) strong additional properties on the curvature of u are required. As in standard demand theory. ρ)) s(w. If beliefs 13 for example: u(µ) := ln(r + µ).

Therefore. V x . Definition 3. then one has to distinguish for each a ∈ A a pair of first and second moment beliefs (q a .3) is closed.1 Given individual second moment beliefs (V a )a∈A and an aggregate stock of assets x ¯ = 0. for given beliefs. V x ¯ a∈A 1 2 (3. This fact is typically suppressed in the standard analysis. If V a = V b the contract curve is a straight line. an asset allocation (xa )a∈A induces a risk allocation (σ a )a∈A by 1 σ a := xa . q a − Rp) = x ¯. It is immediate that the set of feasible risk allocations S := (σ a ) ∈ R+ | σ a := xa . given individ(subjective) variance or standard deviation σ (x) := ual second moment beliefs (V a )a∈A .2) with at least one strict inequality. 20 . the efficient boundary Sef f is given by the simplex Sef f = (σ a ) ∈ R + | |A| σa = x ¯. and convex. Following standard equilibrium terminology an asset allocation (xa )a∈A is said to be feasible if a xa = x ¯. Since individual demand is a function of subjective asset premia q a − Rp. bounded below. an asset market equilibrium is obtained by a vector of equilibrium asset prices such that ϕa (wa . The ”contract curve” connecting the corners a and b is the (the projection) of the set of efficient risk allocations in the space for agent a’ assets. heterogeneous first moment beliefs enter explicitly as an argument (as an additive shift of prices) in each demand function while second moment beliefs (V a )a∈A determine the functional form parametrically. In the context of mean–variance preferences the risk of a portfolio x is measured by its x. if there is no other feasible asset allocation y a . (3. if second moment beliefs are identical.3 Asset Market Equilibrium 20 of consumers are not the same. xa = x ¯ a∈A (3.1 shows the set of feasible allocations in an Edgeworth box representation for two agents and two assets displaying the contours of the two variances σ (xa ) and σ (¯ x − xa ).4) Figure 3. V a y a ≤ σ a for all a ∈ A (3.1) a∈A Any vector of equilibrium asset prices induces a feasible asset allocation (xa )a∈A . a ∈ A such that √ y a . Moreover. Therefore. V a xa 2 . an asset allocation (xa )a∈A is said to induce an efficient risk allocation (σ a )a∈A . |A| 1 a ∈ A. V a )a∈A . V a xa 2 a ∈ A .

V −1 π 1 2 V −1 π = x ¯ (3. 21 .7) Existence of equilibria with heterogeneous beliefs will be discussed in a later section.1) for all consumers14 . one finds that π is an equilibrium premium if and only if the condition sa (wa .1 Existence and Uniqueness of Equilibrium Continuing with the assumption of common (identical) expectations satisfying (2. π ) = x ¯. such that ϕ a (w a .5) In such a case one obtains the following lemma. any asset market equilibrium induces an efficient risk allocation. the equilibrium condition can be rewritten as π ∈ R K if ϕ a (w a .1: Efficient Risk Allocations ¢§£ ¨ If first moment expectations are identical. 3. any equilibrium asset price vector induces the same equilibrium asset premium. Lemma 3.6) a∈A Exploiting the features of the factorization lemma (2.2). π. an asset market equilibrium is achieved for a vector of subjective premia π .1 If first moment beliefs are identical. π ) = x ¯. a∈A (3. (3.3.1 Existence and Uniqueness of Equilibrium 21 ¢§¦¨ ¢¤£¥ ¡   ¢ ¦¥ Figure 3. V −1 π 2 ) a∈A 14 1 1 π.

Therefore. Then. V x ¯ x ¯. V −1 π 1 2 and define aggregate demand for risk as s((wa )a∈A .9) where s−1 ((wa )a∈A . Choose λ > 0 arbitrary and define π := λV x ¯ = 0.1 Existence and Uniqueness of Equilibrium 22 holds. λ = In other words. If. V −1 π 1 2 V −1 π = x ¯ it follows that any equilibrium premium π must be colinear to V x ¯. ρ) → s((w a )a∈A . V x ¯ x ¯. for risk is equal to aggregate risk for its induced shadow return λ x 15 revealing a formal equivalence to the method of proof used by Dana (1999) 22 . Notice that the existence of an equilibrium in the markets for the K assets is reduced to finding a zero of the one dimensional mapping of excess demand for risk15 .8) and associated equilibrium asset prices p= 1 R q− 1 x ¯. V −1 π 1 2 1 2 V −1 π (3. Proof: From the asset market equilibrium equation (3.7) sa (wa .10) ⇐⇒ x ¯= ⇐⇒ s (wa )a∈A . π. Let ρ := π.12) s (wa )a∈A . x ¯. V x ¯ s−1 (wa )a∈A . ρ) Lemma 3. V −1 π 2 ) 1 1 π.11) (3. ((wa )a∈A . demand for risk is strictly monotonic in ρ. V x ¯ . V ) satisfying (2. V x ¯ s−1 (wa )a∈A .1). ·) is the inverse of the aggregate demand function for risk with respect to ρ. ρ) := a∈A sa (wa . ρ) is continuous and surjective in the shadow return ρ. V x ¯ Vx ¯ (3. x ¯. V x ¯ (3. V x ¯ Vx ¯ (3. in addition. V −1 π 2 ) a∈A 1 1 π. ·) : R+ → R+ . there exists an asset market equilibrium for every aggregate asset supply x ¯ = 0 and arbitrary homogeneous beliefs (q. π := λV x ¯ is an equilibrium premium if and only if aggregate demand ¯. π. V x ¯ 1 2 x ¯ x ¯. λ x ¯.3.2 If aggregate demand for risk s((w a )a∈A . x ¯= a∈A sa (wa . then asset equilibrium is unique with an equilibrium premium given by π= 1 x ¯.

1 Existence and Uniqueness of Equilibrium 23 continuity and surjectivity of the left hand side of (3. Assume that preferences of all consumers a ∈ A fulfill assumptions (2. beliefs.1 Let beliefs of consumers be identical and satisfy (2. x ¯.1). there exists a unique asset market equilibrium for any expectations (q. and the Inada conditions (2. The following theorem provides a positive answer for the subclass of separable preferences. s a (w a . Hence.9).8) and for the equilibrium asset price vector (3.1). i. ρ∗ = λ∗ 1 x ¯. x ¯.12) in λ implies existence of an asset market equilibrium.3. e.29). We close this section with two theorems of existence and uniqueness by stating two sets of economic assumptions on preferences.2) it is sufficient to show that aggregate demand for risk is continuous. QED. V x ¯ (3. If risk is a normal good for all consumers and at least one consumer has quasi linear preferences in µ. Therefore. Moreover. V x ¯ is the equilibrium shadow return to risk.41). and wealth to guarantee a unique asset market equilibrium.3). aggregate demand is surjective and there exists a unique shadow return ρ∗ with ρ∗ = s−1 (wa )a∈A . ρ→0 Normality of risk implies that the demand for risk is increasing in ρ. The assumption that risk is a normal good for all consumers seems fairly restrictive. Assumptions (2.13) x ¯. ρ) = 0. (2. strictly monotonic and surjective in ρ. . Strong monotonicity implies the unique solution λ∗ = Thus. V x ¯ s−1 (wa )a∈A . V x ¯. Theorem 3. Proof: In view of lemma (3. QED. V x ¯ . it is desirable to obtain existence and uniqueness in such cases as well. V ) and any aggregate supply of assets x ¯ = 0. implying that the sum of individual risk is equal total risk. λ ∗ a∈A x ¯. V x ¯ )= x ¯.3). This yields the equations for the unique equilibrium premium (3. 23 . the Inada conditions and quasi linearity for some agent a imply that ρ→∞ lim sa (wa . since there is a large set of reasonable preferences for which risk is an inferior good (as in examples (2.32) and (2. ρ) = ∞.41) imply that individual demand for risk exists and is a continuous function for all ρ > 0 with lim sa (wa . and the Inada conditions (2.

Thus. Example 3 provides a case where the monotonicity condition is satisfied but demand for risk is bounded. V x ¯ 1 2 Vx ¯ (3. Thus.17) such that s((w a )a∈A . V x ¯ x ¯= s a (w a . For such a situation it is easy to construct examples with no equilibrium and beliefs satisfying assumptions (2. ρ ¯) = x ¯.3) for all consumers a ∈ A imply that aggregate demand for risk is differentiable. π ¯) = s a (w a . while the monotonicity condition (3.16) Then there exists a unique asset market equilibrium for any expectations (q. aggregate demand for risk satisfies the conditions of lemma (3. and (v a ) (σ ) > 0 for all σ > 0. V x ¯ 2 . This corresponds to a form of the ’beta pricing rule’. V ) and any aggregate supply of assets x ¯ = 0. The condition (3. ρ ¯) x ¯ σ ¯ which is a positive multiple of the market portfolio.16) for at least one agent implies that limρ→∞ sa (wa . ρ) = ∞ which yields this property for the aggregate demand for risk. However.1 Existence and Uniqueness of Equilibrium 24 Theorem 3. asset prices are given by p= 1 ρ ¯ 1 q− R x ¯. σ →∞ lim (v a ) (σ ) = ∞. for preferences with strict concavity in µ.1)16 . and aggregate asset supply.14) (3.2) implies that individual asset demand is given by ϕ a (w a .15) ∆ (ua ) (µ + ∆) strictly increasing in ∆ ≥ 0 for all and for at least one consumer ∆→∞ µ ∈ R. Therefore. QED. equilibrium asset prices are an affine map in expected prices plus an additive constant which depends on preferences. lim ∆ (ua ) (µ + ∆) = ∞ (3.3).1) and (2. ρ ¯) x ¯.55) that the demand for risk is an increasing function for every agent a ∈ A.2 Assume that beliefs of consumers are identical satisfying (2. The factorization lemma (2.15) assures uniqueness only. (3. one has (v a ) (0) = 0. which can be transformed directly into the usual affine relation ship between returns.16) is crucial in guaranteeing existence. In addition. Finally. second moment beliefs. in equilibrium agents hold the same mix of assets as the market portfolio and they share aggregate risk proportionally. (3. If.1) and that preferences satisfy assumption (2.3.2). the two sets of assumptions are mutually exclusive. It follows from (2. since risk is an inferior good under separability. confirming the mutual fund result of the CAPM literature. Observe that equilibrium prices are bounded above by the discounted subjective first 24 . Observe that both of the preceding theorems include in their assumptions the case of quasi linear preferences in µ for all agents. Proof: Assumptions (2. for all consumers a ∈ A.

3.1 Existence and Uniqueness of Equilibrium 25    ¤ ¢ £¥¤ ¦¨§ © .

2: No equilibrium x ˜      ¨  !   Figure 3.3: Two equilibria 25 . ¦ ¡ Figure 3.

1 Existence and Uniqueness of Equilibrium 26   ¢¤£¦¥¦§ ¨ ©  .3.

 .

and (3.3). In the case of piece wise quasi linear preferences it is apparent that the demand for risk cannot be differentiable and monotonically increasing for all wealth levels. 26 .4) provide geometric insight into the existence and uniqueness issue. There can be multiple sections with deceasing demand for risk. (3.15) for positive µ in her proof of existence.4: Three equilibria moment expectations if V x ¯ is non negative. ¡ Figure 3.2). 16 Dana (1999) uses only (3.5. and prices are positive only if expectations are sufficiently optimistic. This implies the possibility of an arbitrary finite number of equilibria. They display the aggregate offer curve for three different situations for which existence or uniqueness fails. Figures (3. a situation depicted in Figure 3.

Mas-Colell 1985)) provide the parallel results for the case here. initial wealth.6).18) If the demand for risk is differentiable and regular. V x ¯ 2 denote the market risk. p) 1 ρ ¯     p = q − V x ¯   1   R x ¯. and (3. Define ρ ¯ > 0 to be the unique market clearing shadow return. q. (3. σ. ρ) ∈ R|A| × R2 + | s((w )a∈A .       (3. Since the asset price equation (3. V ) and x ¯ = 0 be given and σ ¯ := x ¯.19) Figures (3. For a smooth regular aggregate demand function for risk. (3. ρ ¯) = x ¯.2 The equilibrium set and determinacy 27 ¡ £ ¢ ¡   Figure 3. i. s((wa )a∈A . standard results from the theory of regular economies ((cif. e.20) W := (w.5: Multiple equilibria with piece wise quasi linear preferences 3. 1 (3.17) is a smooth mapping. V. 1 Let (q.7). ρ) = σ } . V x ¯ .3. ρ ¯) = x ¯. x ¯.8) show the three essential cases which may occur. the equilibrium manifold for asset prices   1  2 ρ ¯> 0  s(w. equilibria are determinate and generically finite. V x ¯ 2. V x ¯ 2 27 . and aggregate stock of assets.2 The equilibrium set and determinacy In the case of uniqueness we can now summarize the results as follows providing a complete characterization of asset prices under general and homogeneous beliefs for arbitrary preferences. implying the typical structure of the equilibrium manifold for given preferences and second moment beliefs a Wρ := (w.

3.2 The equilibrium set and determinacy 28   .

¡£¢¥¤§¦©¨ Figure 3.6: Unique equilibrium x ˜ .

7: Global equilibrium manifold x ˜ 28 . £¥ Figure 3.

2 The equilibrium set and determinacy 29   ¡£¢ ¤¦¥¨§ ©.3.

such equilibria are mean variance efficient given expectations.1 Let beliefs satisfy (2.17) restricted to the smooth manifold (3. but conceptually quite distinct. Thus. the set of equilibrium asset prices is finite. 29 . Thus. V.8: Bounded equilibrium manifold x ˜ with w := (w a )a∈A and K (w. Then. x ¯. q.19). Figure 3. The fact that beliefs are homogeneous (identical) reduces the existence issue to finding a zero of a one dimensional excess demand function for risk. Proof: The equilibrium manifold W is the graph of the smooth price map (3. QED.1) and assume that aggregate asset demand s(w. one observes that existence and uniqueness of equilibrium are guaranteed under a set of intuitive economic assumptions on preferences and beliefs for the mean variance case. Thus. K W ⊂ R|A| × RK × RK × RK + ×R 2 is a smooth manifold. ρ ¯) is regular. p) ∈ R|A| × RK × RK × RK + ×R 2 is well behaved in general. Summarizing the results so far. At the same time. the following proposition is an immediate consequence of standard results from smooth equilibrium analysis (see for example Mas-Colell (1985)). Proposition 3. formally similar to the reduction used by Dana (1999).

+1 Let (ea . For every agent a ∈ A. the equilibrium shadow return to risk ρ ¯ is the same for all consumers and equal to the subjective Sharpe ratio (i. with total asset endowments x the list {(U a . x ¯ a ) ∈ RK + (4. Then.1). denote agent a’s endowment of the numeraire commodity + a ¯ = Ax e ∈ R+ and of the risky assets x ¯ a ∈ RK ¯a . 30 . Definition 4. V a ) satisfies assumption (2.3) 4. + .2) and. V ).17) as a general price law of P : R+ → RK defined by p = P (ρ) := ρ 1 q− R x ¯. Thus.4 Equilibria with asset endowments 30 in equilibrium. consider the equilibrium asset price equation (3. the relationship between the previous model and the private ownership situation is the usual one.2) +1 (ii) (ea . It is straightforward to extend the above model to the case with individual asset endowments and endogenous wealth. (4.1 A mean–variance asset exchange economy with private ownership E is a list E := {(U a . x ¯a ))})a∈A is an exchange economy in the usual sense. As a consequence. x ¯ a ) ∈ RK .1 Homogeneous Beliefs Definition 4. V a ))a∈A } such that for all a ∈ A: (iii) U a satisfies assumption (2. for given beliefs (q a . V a ). e.1) (i) (q a . For any asset price vector p ∈ RK . (ea . V x ¯ 1 2 Vx ¯ (4. x ¯a . an asset price vector p ¯ ∈ RK is an equilibrium price vector if there exists a shadow return ρ ¯ ≥ 0 and a wealth distribution w ¯ := (w ¯ a )a∈A ∈ R|A| . (q a . the ratio of expected return to standard deviation). individual risk taking induces an efficient (ex ante) risk sharing. define the wealth function W a (p) := ea + p. x ¯a )). 4 Equilibria with asset endowments Consider the capital asset pricing model as a pure exchange model with private ownership of asset endowments. a ∈ A. (ea . In this situation total wealth of agents is endogenous and price dependent.2 Given homogeneous beliefs (q. asset bundles are all positive multiples of the aggregate market portfolio.

The second theorem which allows for risk inferiority parallels that of Theorem (3.9) (4.1) such that q >> 0 and V x ¯ 0.2). and (v a ) (σ ) > 0 for all ∆ (ua ) (µ + ∆) strictly increasing in ∆ ≥ 0 for all and for at least one consumer ∆→∞ µ ∈ R.41). lim ∆ (ua ) (µ + ∆) = ∞ (4.1 and 3. σ →∞ a∈A ea > 0.1) such that q >> 0 and V x ¯ 0.16). Notice that the conditions (4.2 Let individual endowments (ea .3 such that: (v a ) (0) = 0.8). for all σ > 0. x ¯a ) ≥ 0 with x ¯ = a∈A x ¯ a = 0.11) are the same as (3.4.2. ρ ¯) a∈A (4.1 Homogeneous Beliefs 31 such that x ¯ = a∈A 1 2 x ¯a sa (w ¯a. x ¯a In other words an asset market equilibrium is a fixed point of the four mappings given by the right hand sides of equations (4. 31 . Theorem 4.2). Assume that preferences of all consumers a ∈ A fulfill assumptions (2. We only give the proof of the extension for the case of Theorem (3. preferences are separable with 2. a∈A ea > 0.1 Let individual endowments (ea .14)–(3. V x ¯ 1 2 Vx ¯ 1 2 (4.4) x ¯. consumers a ∈ A.6) (4.5)– (4.3). Therefore existence and uniqueness can now be proved exploiting the properties derived in the previous sections. (4. x ¯a ) ≥ 0 with x ¯ = a∈A x ¯a = 0. If risk is a normal good for all consumers and at least one consumer has quasi linear preferences in µ. Theorem 4.8) (q − R p ¯).10) lim (v a ) (σ ) = ∞. V −1 (q − Rp ¯) w ¯ a = W a (¯ p) = ea + p ¯.9)–(4. and the Inada conditions (2. there exists a unique equilibrium asset price p ¯ ∈ RK for the private ownership economy for all beliefs satisfying (2.11) there exists a unique equilibrium asset price p ¯ for the private ownership economy for all beliefs satisfying (2. If. The two results are directly parallel of the two previous theorems 3.7) a∈A (4.5) p ¯ = ρ ¯ = ρ ¯ 1 q− R x ¯. V x ¯ = (4.

V x ¯ 1 2 QED. V x ¯ 1 2 ¯ := w ∈ R|A| | w ≤ wa ≤ w. w := min ea + x a∈A 1 2 = s(w.1 Homogeneous Beliefs 32 Proof: Theorem (3. If ρ1 = r(w1 ) = r(w 2 ) = ρ2 .4. r is the solution function for equation (4. ρ 2 ) so that x ¯. The following example indicates that an assumption on the limiting behavior such as (2. Since H is continuous. Since preferences are assumed to be separable and strictly concave. Suppose ρ1 > ρ2 > 0.56) cannot be dispensed with if existence is to be guaranteed. V x ¯ which is a contradiction. ¯ a∈A W Vx ¯ (4. ρ 1 ) = x ¯. R|A| w → R|A| → H (w ) 1 r(w ¯) q− R x ¯. ρ 2 ) > a s a (w 1 . one finds that s a (w 1 . (4. x ¯.12) H is continuous and each H a (w) is bounded above by w ¯ := and below by ¯a . The assumption (??) implies that the demand for risk increases in the shadow return ρ. implying w 2 > w1 since endowments are positive. This implies H : W ∗ ∗ induces an equilibrium shadow returnρ = r(w ) and equilibrium asset prices p∗ = 1 ρ∗ q− R x ¯. ρ 1 ) > s a (w 2 . ρ 1 ) > s a (w 2 .7). there exists a fixed point w ∗ ∈ W ¯ . risk is an inferior good.2) implies the existence of a continuous function r : R|A| → R+ defining the unique equilibrium shadow return ρ = r(w) for every initial wealth w = (w a )a∈A . i.13) To prove uniqueness. V x ¯ 1 1 2 1 ρ2 Vx ¯ <p = q− R x ¯. This yields 1 ρ1 p = q− R x ¯. r(w)) = a sa (wa . V x ¯ Consider the mapping H := defined by H (w) = (H a (w))a∈A H a (w) := (W a ◦ P ◦ r)(w) ¯. 1 2 = A s a (w 2 . Therefore. V x ¯ 1 2 Hence. ea + x 1 q R Vx ¯ . 32 .13). e. assume to the contrary that there exists two fixed points w 1 = w2 . there exist a unique asset price from equation (4. r(w)). V x ¯ 2 1 2 Vx ¯ . restricting H to the compact convex set ¯ →W ¯ .

These features have an immediate impact on the equilibrium set. Hence. that existence and uniqueness of equilibria are essentially determined by those properties of risk preferences in mean in standard deviation which guarantee globally invertible demand functions for risk. . because of the negative correlation for commodities k ∈ H .5 Conclusion 33 Example Consider an exchange economy with two agents A = {a. (¯ xa )k > 0 k ∈ H. (¯ xb )k = 0 k ∈ H.1 portrays the situation of multiple equilibria with two agents A = {a. U a (µ. when agents hold arbitrary but homogeneous beliefs. b} where b has quasi linear preferences and a a has a piece wise linear utility in the return µ . e b = 0. (4. a (¯ x )k = 0 k ∈ / H.15) (4. Notice that.18) and q >> 0 It is apparent that consumer b will supply (¯ x)b at any price with zero demand for risk. . (¯ x b )k > 0 k ∈ / H. a’s wealth will be larger than ea > 0 implying that his demand for risk will be less than r.14) (4. Finally. . equilibrium prices for these commodities must be positive. . The results of the paper have shown for the CAPM model. the CAPM with mean– variance preferences induces demand behavior explicitly derivable from intuitive assumptions using standard concepts like normality and the Slutzky decomposition. σ )) = ln(r + µ) − σ 2 U b (µ. Therefore. K } such that (V x ¯)k < 0 k ∈ H. there is excess supply of risk for all asset prices. 33 . By treating expectations parameters explicitly the interplay between expectations and actual equilibrium asset prices can be revealed formally. V x ¯ 1 2 Vx ¯ .16) (4. 5 Conclusion Under a general non redundancy assumption on expectations. b} 0<r< x ¯. inducing typically smooth manifolds when preferences are smooth. V x ¯ . since p= 1 R q− λ x ¯.17) (4. a e > 0. For homogeneous expectations globally invertible demand for risk is obtained using elementary techniques. There exists H ⊂ {1. Figure 4. reconfirming the two fundamental findings of CAPM theory: the mutual funds theorem and the beta pricing rule. σ )) = −σ 2 (V x ¯)k > 0 k ∈ / H.

5 Conclusion 34 ¡ ¡¢ § £ ¥¨§ © ¤¦   Figure 4.1: Multiple Equilibria with two agents 34 .

Katzner. (1974): Core and Equilibria of a Large Economy. (1982): “Existence of Competitive Equilibrium”. Intrilligator. Princeton. Econometrica. Econometrica. by K.. & A. 1169– 1174. Journal of Financial and Quantitative Analysis. Chiarella (2000): “Mean Variance Preferences. & H. T. (1985): The Theory of General Economic Equilibrium: A Differentiable Approach. Lintner. Review of Economics and Statistics. Cambridge. (1972): “Smooth Preferences”. & C. North-Holland Publishing Company. 2. ¨ hm. University of Bielefeld. in Handbook of Mathematical Economics II. Hildenbrand. ¨ hm. V. 36. 59. 23. 603–615. V. Nielsen. New York a. Uniqueness and Determinacy of Equilibrium in CAPM with a Riskless Asset”. H. S. Journal of Mathematical Economics. Arrow & M. 233–246.. 37. Laitenberger & A. Dana. 165–173. by K. 2. vol. A. 40. 329–336. Konno. 415–418. Expectations FormaBo tions. Mas-Colell. Econometrica. Arrow & M. Cambridge University Press. J. J. Shirakawa (1995): “Existence of a nonnegative equilibrium price vector in the mean-variance capital market”. D. (1991): “Existence theorems in the CAPM”. Zeitschrift f¨ ur National¨ okonomie. V. North-Holland Publishing Company. Discussion Paper No. LeRoy. Amsterdam a. Intrilligator. Lo of Equilibria in the CAPM”. & J. Debreu. 13–37. 448. (1999): “Existence. 268–290. UK. Werner (2001): Principles of Financial Economics. ed. Mu ¨ ller (1977): “Two Examples of Equilibria under Price Rigidities Bo and Quantity Rationing”. (1968): “A Note on the Differentiability of Consumer Demand Functions”. Journal of Mathematical Economics. chap.o. 167–175. G. (1959): Theory of Value. Barten (1982): “Consumer Theory”. Econometrica. ¨ ffler (2002): “Two Remarks on the Uniqueness Hens. 47. Cambridge (Mass. A. in Handbook of Mathematical Bo Economics. Inc. L. and the Dynamics of Random Asset Prices”. 37. Mathematical Finance. 15. R. 35 . Amsterdam a. (1988): “Uniqueness of Equilibrium in the Classical Capital Asset Pricing Model”. Cambridge University Press.o. 32(2). M. forthcoming in Mathematical Finance. W.REFERENCES 35 References Allingham. ¨ hm. (1965): “The Valuation of Risky Assets and Selection of Risky Investment in Stock Portfolios and Capital Budgets”. vol. (1966): “Equilibrium in a Capital Asset Market”. John Wiley & Sons. 123–132. 3. Princeton University Press.) a. Mossin. ed. J. W.o. 34. & H.o. J.

Journal of Finance.REFERENCES 36 Pliska. Sharpe. Massachusetts. Blackwell Publishers Inc. Annals of Economics and Finance. (1987): “Arbitrage and the Existence of Competitive Equilibrium”. W. 36. S. Sun. Werner. (1997): Introduction to Mathematical Finance. Stapleton. 36 . Econometrica. & Z. Econometrica. 215–235. Subrahmanyam (1978): “A Multiperiod Equilibrium Asset Pricing Model”. 19. (1964): “Capital Asset Pricing: A Theory of Market Equilibrium Under Conditions of Risk”. R. 425–442. & M. 1077–1096. Journal of Economic Theory. 4. 46. 110–119. 55(6). N. F. J. Yang (2003): “Existence of Equilibrium and Zero-Beta Pricing Formula in the Capital Asset Pricing Model with Heterogeneous Beliefs”. 1403–1418. (1985): “Equilibrium in Economies with Incomplete Financial Markets”.