Incomplete-market CAPM equilibria in continuous time

arXiv:2408.03134 · q-fin.MF, econ.GN, q-fin.EC · Submitted 2024-08-06 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Quantitative Finance. The day in quantitative finance: models of markets, risk and prices.

Ruth: We're looking at the paper: "Incomplete-market CAPM equilibria in continuous time".

Warren: Here's the gist of the paper. Existence and uniqueness of quadratic and linear mean–variance equilibria in general semimartingale markets is established by revisiting classical mean–variance equilibria with both financial and real assets, providing necessary and sufficient conditions for existence and uniqueness under general semimartingales in discrete as well as continuous time.

Ruth: Market Setup and Admissible Strategies The paper sets up a filtered probability space for a financial market consisting of numeraire, financial assets, and productive assets.

Warren: It defines the price processes based on local martingale parts and terminal conditions for real assets.

Ruth: Admissible trading strategies are defined using a two-step approach involving simple integrands and an equivalence relation based on indistinguishable stochastic integrals.

Title and authors: Ruth: Let's start by looking at the title and who wrote this: "Incomplete-market CAPM equilibria in continuous time." The key thing is that they are revisiting the classical mean-variance topic but broadening it to include real assets and more general market dynamics.

Warren: It’s written by Christoph Czichowsky, Martin Herdegen, and David Martins. I don't know much about their specific background, but this paper is clearly pushing the boundaries of what we thought was possible for equilibrium pricing under uncertainty.

Iris: From an econometric standpoint, what’s interesting is that they are aiming to close a gap in the literature because most previous work only looked at one period models, while this addresses multi-period and continuous time models.

Ruth: That's right. They are trying to find a rigorous study of existence and uniqueness for CAPM equilibria in multi-period and continuous time settings, which is something that’s been missing in the literature for a long time.

Warren: So they aren't just re-deriving old results; they are building a new framework to see if these equilibrium markets can even exist when you allow for more complex information flows driven by general semimartingales, as mentioned on page two of this paper <ref:2408.03134#pg3>.

Iris: And what that implies is that we’re moving beyond the standard assumptions of Brownian motion, which is crucial because real-world asset returns often exhibit more jumps and discontinuities than a simple continuous process allows.

Ruth: It suggests that the structure of an equilibrium market depends heavily on these underlying mathematical properties of the price processes, not just the standard assumptions we use every day.

The paper's summary: Warren: So, if we look at what they actually do in this paper, it’s that for agents with quadratic utility functions, they provide necessary and sufficient conditions for existence and uniqueness of equilibria, which is the main result in Theorem three point seven <ref:2408.03134#pg3,result in Theorem 3.7>.

Iris: That theorem links the individual optimality problem to a mean-variance hedging problem where the payoff is defined by gamma sub k minus xi sub k, which simplifies things mathematically for the analysis.

Ruth: They show that this quadratic utility problem is equivalent to a mean-variance hedging problem with that specific payoff structure, and their representative agent's maximization structure mirrors the individual one in a way that helps establish this equivalence.

Warren: But then they move on to the harder case, which is linear mean-variance preferences, showing that under mild assumptions, a linear mean-variance equilibrium corresponds to a quadratic equilibrium for different risk parameters.

Iris: They use this link between the two types of preferences to study a fixed-point problem that establishes existence and uniqueness for those linear mean-variance equilibria in Theorem four point one four <ref:2408.03134#pg3>.

Ruth: So, the summary is that they use the connection between hedging problems and utility maximization to prove that these complex equilibrium markets can exist under specific conditions derived from their model primitives on page three of this paper <ref:2408.03134#pg2>.

The paper's improvements: Iris: One of the main improvements they suggest is providing explicit examples showing when equilibria are not unique or even don't exist, which gives us a clearer picture of the boundaries for these models.

Warren: They also explicitly characterize the mean-variance efficient frontier using a triplet involving l, ck, and epsilon two k; that helps us map out exactly how risk tolerance parameters relate to optimal trading strategies in those linear mean-variance preferences.

Ruth: And they provide an explicit solution for the unique optimal strategy in those linear mean-variance preferences as theta sub k lambda k over l, which gives us a direct computational path for finding equilibrium strategies when certain market assumptions hold.

Iris: They also established necessary and sufficient conditions for the existence of quadratic equilibrium markets based on martingale measures, specifically requiring things like z bar t not being zero or z bar t minus one not being zero for all time P-a.s.

Warren: These conditions are important because they allow us to verify market stability and pricing feasibility under structural constraints, like when the buy-and-hold strategy is inadmissible in an L two sense on page three of this paper <ref:2408.03134#pg2>.

Conclusion: Ruth: So to wrap up, the paper "Incomplete-market CAPM equilibria in continuous time" shows that existence and uniqueness are possible for these complex markets if we stick to certain conditions, like the aggregate risk tolerance PK sub k equals one lambda sub k being larger than the uncertainty of the aggregate endowment.

Warren: That condition is really important because it sets a clear threshold: if that inequality doesn't hold, you won't find an equilibrium market of that form. The paper concludes that equilibria with aggregate risk tolerance greater than gamma zero are the most meaningful from an economic point of view.

Iris: I just want to say that the main limitation they state is relying on fine properties of dynamic processes and the need for integrability conditions to be satisfied, which means their results are conditional on those technical assumptions holding true.

Ruth: So, they give us a solid mathematical foundation for understanding when these equilibria exist in continuous time settings with general semimartingales, even if we have to meet some pretty strict technical hurdles.

Warren: This work gives us a much more robust way to think about how asset pricing works when the market dynamics are messy and include real assets, which is something that’s been missing in the literature for years.

London School of Economics and Political Science · University of Warwick · ETH Zurich

q-fin.MF, econ.GN, q-fin.EC

Submitted: 2024-08-06

Updated: 2026-11-19

License: http://creativecommons.org/licenses/by/4.0/

Importance score: 78/100

The gist: Existence and uniqueness of quadratic and linear mean–variance equilibria in general semimartingale markets is established by revisiting classical mean–variance equilibria with both financial and…

Key concepts

Semimartingale Markets
These are advanced financial models where asset price processes are driven by general stochastic processes, not just standard Brownian motion. This allows the model to accurately reflect more realistic trading environments involving both financial and real assets in continuous or discrete time.
Quadratic Utility Analysis
This approach analyzes agent behavior using quadratic utility functions, which is a common assumption for risk aversion. The paper shows that the individual optimization problem simplifies into a mean-variance hedging problem, linking individual choices to market equilibrium.
Mean-Variance Hedging (MVH)
MVH is the core mathematical framework used to find optimal trading strategies under quadratic utility. It transforms the agent's utility maximization into a solvable hedging problem, helping determine if a stable market price exists for assets.

Terminology

Summary

Existence and uniqueness of quadratic and linear mean–variance equilibria in general semimartingale markets is established by revisiting classical mean–variance equilibria with both financial and real assets, providing necessary and sufficient conditions for existence and uniqueness under general semimartingales in discrete as well as continuous time. p. 0

Where the paper says it

August 5, 2024 Abstract We revisit the classical topic of quadratic and linear mean–variance equilibria with both financial and real assets. The novelty of our results is that they are the first allowing for equilibrium prices driven by general semimartingales and hold in discrete as well as continuous time.

Market Setup and Admissible Strategies

The paper sets up a filtered probability space for a financial market consisting of numeraire, financial assets, and productive assets. It defines the price processes based on local martingale parts and terminal conditions for real assets. p. 2

Where the paper says it

1 In addition, we consider d1 financial assets with price processes S (1) = (S 1 t , . , Sd1 t )0≤t≤T and d2 productive assets (sometimes also referred to as real assets) with price processes S (2) = (S d1+1 t , .

Admissible trading strategies are defined using a two-step approach involving simple integrands and an equivalence relation based on indistinguishable stochastic integrals. p. 3
Where the paper says it

2 Admissible strategies In order to describe trading in the market (1, S), we need to specify which strategies are considered admissible. To this end, we follow a two-step approach as in Cern´y/Kallsen [5], to which ˇ we also refer for further details.

The buy-and-hold strategy is denoted by (e j t) and its admissibility depends on the asset being an L2-semimartingale. p. 3
Where the paper says it

, 0) ∈ R d the buy-and-hold strategy of the j-th risky asset, where 1 is in the j-th position. In general, this strategy will not be L 2 -admissible for S because we only assume S j to be a local L 2 -semimartingale.

Quadratic Utility Analysis

For agents with quadratic utility functions, the existence and uniqueness of equilibria are studied through a nonstandard type of representative agent. The individual optimality problem (3. p. 6

Where the paper says it

1 Individual optimality Throughout this section, we fix a local L 2 -market (1, S) that satisfies (2. In order to study the maximisation problem (3.

1) for agent k is shown to be equivalent to a mean–variance hedging problem (3. p. 0
Where the paper says it

We then use this link to study a fixed-point problem that establishes existence (and uniqueness in a suitable class) of linear mean–variance equilibria. Our results rely on fine properties of dynamic mean–variance hedging in general semimartingale markets.

2). A key result shows that the quadratic utility problem is equivalent to an MVH problem with payoff H k:= γ k − Ξ k. p. 7
Where the paper says it

The following result shows that the quadratic utility problem (3. 1) is equivalent to a MVH problem (3.

The representative agent's maximization problem (3.8) is shown to have a structure identical to the individual one. p. 1
Where the paper says it

This structure breaks down in the incomplete case as shown by Koch-Medina/Wenzelburger [17]. They find that in an incomplete market, each agent still hedges their individual endowment as best as possible, even though this cannot be done perfectly.

Equilibrium Conditions for Quadratic Markets

Theorem 3.7 provides necessary and sufficient conditions for the existence and uniqueness of a quadratic equilibrium market under the assumption that Z¯ and Z¯− do not hit 0. p. 0

Where the paper says it

The novelty of our results is that they are the first allowing for equilibrium prices driven by general semimartingales and hold in discrete as well as continuous time. For agents with quadratic utility functions, we provide necessary and sufficient conditions for the existence and uniqueness of equilibria.

The resulting equilibrium price processes are explicitly given by formulas (3. p. 2
Where the paper says it

In the following, we likewise use the notation x = (x (1), x(2)) for each x ∈ R d1+d2 with x (i) ∈ R di . We assume that the price processes S (1) and S (2) are not given a priori but rather determined in equilibrium between K agents trading in the market.

14) and (3.15). Sufficient conditions for the existence of such an equilibrium include Dj ∈ L∞(P) or H, ¯H¯ − 1 ∈ L∞(P). p. 0
Where the paper says it

The novelty of our results is that they are the first allowing for equilibrium prices driven by general semimartingales and hold in discrete as well as continuous time. For agents with quadratic utility functions, we provide necessary and sufficient conditions for the existence and uniqueness of equilibria.

A counterexample is provided illustrating how integrability issues can preclude the existence of an equilibrium. p. 1
Where the paper says it

In addition, one needs to impose integrability conditions on the admissible trading strategies. These conditions can preclude the existence of an equilibrium.

Linear Mean–Variance Equilibria

The analysis extends to linear mean–variance preferences, showing that under mild assumptions, a linear mean–variance equilibrium corresponds to a quadratic equilibrium for different risk parameters. The individual linear mean–variance problem (4. p. 0

Where the paper says it

We first show that under mild assumptions, a linear mean–variance equilibrium corresponds to a quadratic equilibrium (for different preference parameters). We then use this link to study a fixed-point problem that establishes existence (and uniqueness in a suitable class) of linear mean–variance equilibria.

1) is shown to be connected to the quadratic utility problem via mean–variance efficient strategies. p. 1
Where the paper says it

We also provide examples that illustrate when CAPM equilibria are not unique or fail to exist. Because “mean–variance efficient portfolios approximately maximise expected utility for a wide range of risk-averse (concave) utility functions” (see Markowitz [21, 22]), our results may also provide approximations to equilibrium prices for other utility functions.

The mean–variance efficient frontier E k is characterized by a formula involving the triplet (l, ck, ε 2k). p. 1
Where the paper says it

We also provide examples that illustrate when CAPM equilibria are not unique or fail to exist. Because “mean–variance efficient portfolios approximately maximise expected utility for a wide range of risk-averse (concave) utility functions” (see Markowitz [21, 22]), our results may also provide approximations to equilibrium prices for other utility functions.

Existence of Linear Mean–Variance Equilibria

Theorem 4.14 establishes the existence and uniqueness of a linear mean–variance equilibrium of the form (1, S(¯γ)) under Assumption 4. p. 0

Where the paper says it

We first show that under mild assumptions, a linear mean–variance equilibrium corresponds to a quadratic equilibrium (for different preference parameters). We then use this link to study a fixed-point problem that establishes existence (and uniqueness in a suitable class) of linear mean–variance equilibria.

8. The condition for this existence is that the aggregate risk tolerance PK k=1 λ k must be larger than the uncertainty of the aggregate endowment Ξ, as measured by ess sup Ξ¯ − EP [Ξ] ¯ ≥ 0. p. 1
Where the paper says it

Because “mean–variance efficient portfolios approximately maximise expected utility for a wide range of risk-averse (concave) utility functions” (see Markowitz [21, 22]), our results may also provide approximations to equilibrium prices for other utility functions. In most of the extant literature on the existence and uniqueness of CAPM equilibria, it is assumed that the aggregate random endowment is tradable so that the financial market is complete.

This condition is equivalent to X K k=1 λ k > ess sup Ξ¯ − EP [Ξ] ¯ ≥ 0. p. 3
Where the paper says it

In this case, we write ϑ =S ϑ ′ ; see Czichowsky/Schweizer [9] for more details on how to represent different equivalent classes via the so-called projection onto the predictable range. We assume that trading is frictionless and that market participants choose self-financing portfolios of the form (ϑ 0 t , ϑt)0≤t≤T , where ϑ 0 is a predictable process, ϑ ∈ Θ(S) and the wealth process (Vt(ϑ 0 , ϑ))0≤t≤T satisfies the self-financing condition Vt(ϑ 0 , ϑ) = ϑ 0 t + ϑ ⊤ t St = ϑ 0 0 + ϑ ⊤ 0 S0 + ϑ • St , P-a.

Uniqueness and Characterization

The paper demonstrates that for quadratic equilibria, the market clears if and only if the representative agent does not trade, leading to a pricing measure. For linear mean–variance equilibria, the equilibrium price process S(¯γ) is unique for any choice of parameters γ 1,. p. 0

Where the paper says it

We first show that under mild assumptions, a linear mean–variance equilibrium corresponds to a quadratic equilibrium (for different preference parameters). We then use this link to study a fixed-point problem that establishes existence (and uniqueness in a suitable class) of linear mean–variance equilibria.

.., γ K such that PK k=1 γ k = ¯γ. The final result shows that the market (1, S(¯γ)) is a mean–variance equilibrium if and only if the aggregate risk tolerance PK k=1 λ k is larger than the uncertainty of the aggregate endowment Ξ, as measured by ess sup Ξ¯ − EP [Ξ] ¯ ≥ 0. p. 1
Where the paper says it

Because “mean–variance efficient portfolios approximately maximise expected utility for a wide range of risk-averse (concave) utility functions” (see Markowitz [21, 22]), our results may also provide approximations to equilibrium prices for other utility functions. In most of the extant literature on the existence and uniqueness of CAPM equilibria, it is assumed that the aggregate random endowment is tradable so that the financial market is complete.

Key Results on Mean–Variance Hedging

The appendix provides key results on mean–variance hedging problems (3.2) and (3.3). Proposition A. p. 7

Where the paper says it

A sufficient condition for the existence and uniqueness of solutions to (3. 3) is given in Proposition A.

p. 7
Where the paper says it

A sufficient condition for the existence and uniqueness of solutions to (3. 3) is given in Proposition A.

7 shows that if Zt 6= 0 P-a. p. 7
Where the paper says it

We refer to Schweizer [28] for a recent overview of mean–variance hedging. The following result shows that the quadratic utility problem (3.

p. 7
Where the paper says it

We refer to Schweizer [28] for a recent overview of mean–variance hedging. The following result shows that the quadratic utility problem (3.

s., then the MVH problem (3. p. 6
Where the paper says it

In order to study the maximisation problem (3. 1), we note that it is closely linked to the so-called mean–variance hedging problems for a payoff H ∈ L 2 : • The mean–variance hedging (MVH) problem is given by E (ϑ • ST − H) 2 −→ min ϑ∈Θ(S) !

2) and exMVH problem (3. p. 6
Where the paper says it

1), we note that it is closely linked to the so-called mean–variance hedging problems for a payoff H ∈ L 2 : • The mean–variance hedging (MVH) problem is given by E (ϑ • ST − H) 2 −→ min ϑ∈Θ(S) ! 2) • The extended mean–variance hedging (exMVH) problem is given by E (c + ϑ • ST − H) 2 −→ min (c,ϑ)∈R×Θ(S) !

3) have unique solutions for each H ∈ L 2. p. 7
Where the paper says it

We say that (3. 2) has a unique solution if ϑ 1 =S ϑ 2 for any two solutions ϑ 1 , ϑ2 ∈ Θ(S).

The uniqueness of value processes implies the uniqueness of gains processes, which in turn ensures a unique solution to the MVH problem for any payoff H. Lemma 4.6 characterizes the optimal strategy for (4. p. 6
Where the paper says it

It ensures that each buy-and-hold strategy η k is L 2 -admissible, which is a natural requirement since the agents should be allowed to simply hold their respective traded endowments. Mathematically, it also ensures that ϑˆk ∈ Θ(S) if and only if ϑˆk − η k ∈ Θ(S), which will allow us to find the optimal strategies ϑˆk by first solving for ϑˆk − η k ; recall also the discussion below (2.

1) as ϑ k(λ k/l). The relationship between the mean–variance equilibrium and quadratic equilibria is established by showing that any linear mean–variance equilibrium must be a quadratic equilibrium with specific risk tolerances γ k:= ck + λ k/l. p. 1
Where the paper says it

To the best of our knowledge, this is the first result that allows for equilibrium prices driven by general semimartingales in an incomplete setting. In the case of linear mean–variance preferences, we combine the above result with a fixedpoint approach, showing first that under mild assumptions on the model primitives, a linear mean–variance equilibrium corresponds to a quadratic equilibrium for different risk parameters.

Dynamic Theory and Fixed Points

The dynamic theory introduces the opportunity process L(¯γ) as the value process of the pure investment problem. The mean value process V t(¯γ) is defined in terms of L t(¯γ) and a semimartingale term, satisfying V T(¯γ) = H¯ (¯γ). p. 4

Where the paper says it

Thus the terminal wealth of agent k at time T can be equivalently written as V k T (ϑ) = Ξk,tr − η k • ST + ϑ • ST + Ξk,n = (ϑ − η k ) • ST + Ξk (2. 6) in terms of the total endowment defined in (2.

The fixed-point condition for mean–variance equilibria is formulated as γ̄ = ˜γ(¯γ), which is solved by the relationship γ̄ − EP [Ξ] = ¯c(¯γ) / l(¯γ). p. 1
Where the paper says it

In the case of linear mean–variance preferences, we combine the above result with a fixedpoint approach, showing first that under mild assumptions on the model primitives, a linear mean–variance equilibrium corresponds to a quadratic equilibrium for different risk parameters. Due to the linearity assumption in the mean–variance preferences and with the help of the general theory on mean–variance hedging developed in the early 2000s, see Cern´y/Kallsen [5], ˇ the fixed-point problem can be solved explicitly.

This fixed-point equation is solved to find the unique mean–variance equilibrium of the form (1, S(¯γ)). p. 1
Where the paper says it

In the case of linear mean–variance preferences, we combine the above result with a fixedpoint approach, showing first that under mild assumptions on the model primitives, a linear mean–variance equilibrium corresponds to a quadratic equilibrium for different risk parameters. Due to the linearity assumption in the mean–variance preferences and with the help of the general theory on mean–variance hedging developed in the early 2000s, see Cern´y/Kallsen [5], ˇ the fixed-point problem can be solved explicitly.

The final theorem confirms that this solution exists if and only if ¯γ > γ¯0. p. 0
Where the paper says it

, functionals of the form U(µ, σ), where U is quasiconcave, increasing in the mean µ and decreasing in the volatility σ. This is because without distributional assumptions on the returns, the only utility functions that are compatible with the CAPM (more precisely, the two-fund separation theorem) are quadratic utility functions; see the discussion in Berk [2, after Corollary 3.

Conclusion on Mean–Variance Equilibria

Theorem 4.14 states that the market (1, S(¯γ)) is the unique linear mean–variance equilibrium of the form (1, S(¯γ')) for some γ̄' > γ¯0. The condition for this uniqueness is that ˜γ = ¯γ. p. 7

Where the paper says it

3) has a unique solution if c1 = c2 and ϑ 1 =S ϑ 2 for any two solutions (c1, ϑ1 ),(c2, ϑ2 ) ∈ R×Θ(S). A sufficient condition for the existence and uniqueness of solutions to (3.

This implies that the market (1, S(¯γ)) is a mean–variance equilibrium if and only if PK k=1 λ k > ess sup Ξ¯ − EP [Ξ] ¯ ≥ 0. p. 1
Where the paper says it

In most of the extant literature on the existence and uniqueness of CAPM equilibria, it is assumed that the aggregate random endowment is tradable so that the financial market is complete. The completeness assumption implies the existence of a so-called representative agent, which simplifies the task of showing existence and uniqueness of an equilibrium: Every agent chooses to hedge their idiosyncratic risk and own a fraction of the market portfolio.

The study concludes that equilibria with aggregate risk tolerance ¯γ > γ¯0 are the most meaningful from an economic point of view. p. 5
Where the paper says it

8) where U Q k (x) := 2γkx − x 2 for some risk-tolerance γk ∈ R. The parameter γk is also sometimes called the bliss point of agent k, as it is the optimal wealth that the agent would like to attain at time T in order to maximise their utility.

References

[1] Michael Allingham, Existence theorems in the capital asset pricing model, Econometrica 59 (1991), 1169–1174.

[2] Jonathan Berk, Necessary conditions for the CAPM, Economic Theory 73 (1997), 245–257. p. 0

Where the paper says it

Despite its limitations, it is still one of the cornerstones of modern financial theory and widely used in investment practice; see [18] for a recent overview. Notwithstanding the enormous influence of the CAPM, a rigorous study of existence and uniqueness of CAPM equilibria was initiated only around 1990 by Nielsen [25, 26, 27] and Allingham [1], with more recent important contributions by Berk [2], Dana [11], Hens et al.

[3] D. T. Breeden, An intertemporal asset pricing model with stochastic consumption and investment opportunities, Journal of financial Economics 7 (1979), no. p. 0

Where the paper says it

1 Introduction The capital asset pricing model (CAPM) of Treynor [32], Sharpe [29], Lintner [19, 20] and Mossin [24] is one of the first general equilibrium models for financial markets. Despite its limitations, it is still one of the cornerstones of modern financial theory and widely used in investment practice; see [18] for a recent overview.

3, 265–296.

[4] Aleˇs Cern´y and Christoph Czichowsky, The law of one price in quadratic hedging and mean–variance portfolio selection, Preprint, City, University of London, (v2) available at arXiv: https://arxiv.org/abs/2210.15613, 2024. p. 0

Where the paper says it

03134v1 [q-fin. MF] 6 Aug 2024 Existence and uniqueness of quadratic and linear mean–variance equilibria in general semimartingale markets Christoph Czichowsky∗ Martin Herdegen† David Martins ‡ .

[5] Aleˇs Cern´y and Jan Kallsen, On the structure of general mean–variance hedging strategies, Annals of Probability 35 (2007), 1479–1531. p. 1

Where the paper says it

In the case of linear mean–variance preferences, we combine the above result with a fixedpoint approach, showing first that under mild assumptions on the model primitives, a linear mean–variance equilibrium corresponds to a quadratic equilibrium for different risk parameters. Due to the linearity assumption in the mean–variance preferences and with the help of the general theory on mean–variance hedging developed in the early 2000s, see Cern´y/Kallsen [5], ˇ the fixed-point problem can be solved explicitly.

[6], A counterexample concerning the variance-optimal martingale measure, Mathematical Finance 18 (2008), 305–316. p. 1

Where the paper says it

The aim of this paper is to close this gap in the literature and to address the important question of existence and uniqueness of CAPM equilibria in multiperiod and continuous time models. By exploiting the connection of mean–variance portfolio selection with the classical problem of mean–variance hedging in mathematical finance, see Schweizer [28] for an overview, we provide existence and uniqueness results for CAPM equilibria.

[7] Alexander Cherny and Albert Shiryaev, Vector stochastic integrals and the fundamental theorem of asset pricing, Tr. Mat. Inst. Steklova (Stokhast. Finans. Mat.) 237 (2002), 12–56.

[8] Tahir Choulli, Leszek Krawczyk, and Christophe Stricker, E-martingales and their applications in mathematical finance, Annals of Probability 26 (1998), 853–876. p. 1

Where the paper says it

The aim of this paper is to close this gap in the literature and to address the important question of existence and uniqueness of CAPM equilibria in multiperiod and continuous time models. By exploiting the connection of mean–variance portfolio selection with the classical problem of mean–variance hedging in mathematical finance, see Schweizer [28] for an overview, we provide existence and uniqueness results for CAPM equilibria.

[9] Christoph Czichowsky and Martin Schweizer, Closedness in the semimartingale topology for spaces of stochastic integrals with constrained integrands, S´eminaire de Probabilit´es XLIII (Catherine Donati-Martin, Antoine Lejay, and Alain Rouault, eds. p. 0

Where the paper says it

03134v1 [q-fin. MF] 6 Aug 2024 Existence and uniqueness of quadratic and linear mean–variance equilibria in general semimartingale markets Christoph Czichowsky∗ Martin Herdegen† David Martins ‡ .

), Springer, 2011.

[10], Cone-constrained continuous-time Markowitz problems, Annals of Applied Probability 23 (2013), 764–810. p. 2

Where the paper says it

2 Model and preliminary results 2. 1 Financial market We work on a filtered probability space (Ω, F, F = (Ft)0≤t≤T , P) with a fixed finite time horizon T ∈ (0,∞).

[11] Rose-Anne Dana, Existence, uniqueness and determinacy of equilibrium in C. p. 0

Where the paper says it

Despite its limitations, it is still one of the cornerstones of modern financial theory and widely used in investment practice; see [18] for a recent overview. Notwithstanding the enormous influence of the CAPM, a rigorous study of existence and uniqueness of CAPM equilibria was initiated only around 1990 by Nielsen [25, 26, 27] and Allingham [1], with more recent important contributions by Berk [2], Dana [11], Hens et al.

A.P.M. with a riskless asset, Mathematical Economics 32 (1999), 167–175. p. 4
Where the paper says it

, K} interacts with the market by buying and selling assets according to an L 2 -admissible strategy ϑ ∈ Θ(S), which includes the original endowment η k,(2) in the productive assets. Since the agent does not own any riskless or financial assets at time 0, their initial wealth is η k,(2)⊤ S (2) 0 , which is the initial value of their traded endowment.

[12] Freddy Delbaen and Walter Schachermayer, Attainable claims with p’th moments, Annales de l’IHP Probabilit´es et Statistiques 32 (1996), 743–763. p. 3

Where the paper says it

see Delbaen/Schachermayer [12] and Cern´y/Kallsen [5] for details. We refer to this property ˇ by calling (1, S) a local L 2 -market.

[13], A simple counterexample to several problems in the theory of asset pricing, Mathematical Finance 8 (1998), 1–11. p. 0

Where the paper says it

1 Introduction The capital asset pricing model (CAPM) of Treynor [32], Sharpe [29], Lintner [19, 20] and Mossin [24] is one of the first general equilibrium models for financial markets. Despite its limitations, it is still one of the cornerstones of modern financial theory and widely used in investment practice; see [18] for a recent overview.

[14] Ernst Eberlein and Jan Kallsen, Mathematical finance, Springer, 2019. p. 1

Where the paper says it

The aim of this paper is to close this gap in the literature and to address the important question of existence and uniqueness of CAPM equilibria in multiperiod and continuous time models. By exploiting the connection of mean–variance portfolio selection with the classical problem of mean–variance hedging in mathematical finance, see Schweizer [28] for an overview, we provide existence and uniqueness results for CAPM equilibria.

[15] Thorsten Hens, J¨org Laitenberger, and Andreas L¨offler, Two remarks on the uniqueness of equilibria in the CAPM, Mathematical Economics 37 (2002), 123–132. p. 0

Where the paper says it

This is because without distributional assumptions on the returns, the only utility functions that are compatible with the CAPM (more precisely, the two-fund separation theorem) are quadratic utility functions; see the discussion in Berk [2, after Corollary 3. For expected quadratic utility, existence and uniqueness of CAPM ∗London School of Economics and Political Science, Department of Mathematics, Columbia House, Houghton Street, London WC2A 2AE, UK, email: c.

[16] Jean Jacod and Albert Shiryaev, Limit theorems for stochastic processes, 2nd ed. p. 3

Where the paper says it

We denote by Θsimple(S) the linear space of all simple integrands for S. We also let L(S) be the set of predictable S-integrable processes on [0, T]; see Jacod/Shiryaev [16, III.

, Springer, 2003.

[17] Pablo Koch-Medina and Jan Wenzelburger, Equilibria in the CAPM with non-tradeable endowments, Mathematical Economics 75 (2018), 93–107. p. 0

Where the paper says it

Notwithstanding the enormous influence of the CAPM, a rigorous study of existence and uniqueness of CAPM equilibria was initiated only around 1990 by Nielsen [25, 26, 27] and Allingham [1], with more recent important contributions by Berk [2], Dana [11], Hens et al. [15], Wenzelburger [33] and Koch-Medina/Wenzelburger [17].

[18] Haim Levy, The capital asset pricing model in the 21st century: Analytical, empirical, and behavioral perspectives, Cambridge University Press, 2011. p. 0

Where the paper says it

Our results rely on fine properties of dynamic mean–variance hedging in general semimartingale markets. 1 Introduction The capital asset pricing model (CAPM) of Treynor [32], Sharpe [29], Lintner [19, 20] and Mossin [24] is one of the first general equilibrium models for financial markets.

[19] John Lintner, Security prices, risk, and maximal gains from diversification, Finance 20 (1965), 587–615.

[20], The valuation of risky assets and the selection of risky investments in stock portfolios and capital budgets, Review of Economic Statistics 47 (1965), 346–382. p. 2

Where the paper says it

, Sd1+d2 t )0≤t≤T . The risky assets are collectively expressed as S := (S (1), S(2)).

[21] Harry Markowitz, Portfolio theory: as i still see it, Annu. p. 0

Where the paper says it

1 Introduction The capital asset pricing model (CAPM) of Treynor [32], Sharpe [29], Lintner [19, 20] and Mossin [24] is one of the first general equilibrium models for financial markets. Despite its limitations, it is still one of the cornerstones of modern financial theory and widely used in investment practice; see [18] for a recent overview.

Rev. Financ. Econ. 2 (2010), no. 1, 1–23.

[22], Mean–variance approximations to expected utility, European Journal of Operational Research 234 (2014), no. p. 1

Where the paper says it

We also provide examples that illustrate when CAPM equilibria are not unique or fail to exist. Because “mean–variance efficient portfolios approximately maximise expected utility for a wide range of risk-averse (concave) utility functions” (see Markowitz [21, 22]), our results may also provide approximations to equilibrium prices for other utility functions.

2, 346–355.

[28] Martin Schweizer, Mean–variance hedging, Encyclopedia of Quantitative Finance (Rama Cont, ed. p. 1

Where the paper says it

The aim of this paper is to close this gap in the literature and to address the important question of existence and uniqueness of CAPM equilibria in multiperiod and continuous time models. By exploiting the connection of mean–variance portfolio selection with the classical problem of mean–variance hedging in mathematical finance, see Schweizer [28] for an overview, we provide existence and uniqueness results for CAPM equilibria.

), Wiley, 2010.

[29] William Sharpe, Capital asset prices: A theory of market equilibrium under conditions of risk, Finance 19 (1964), 425–442. p. 0

Where the paper says it

1 Introduction The capital asset pricing model (CAPM) of Treynor [32], Sharpe [29], Lintner [19, 20] and Mossin [24] is one of the first general equilibrium models for financial markets. Despite its limitations, it is still one of the cornerstones of modern financial theory and widely used in investment practice; see [18] for a recent overview.

[30] R. C. Stapleton and M. p. 1

Where the paper says it

While generalisations of the CAPM to multi-period and continuous time models have been considered from the late 1970s onwards (see e. Stapleton/Subrahmanyam [30] and Breeden [3]), a rigorous study of the existence and uniqueness of CAPM equilibria in multi-period and continuous time models is still missing in the literature — all the papers cited in the previous paragraph study one-period models.

G. Subrahmanyam, A multiperiod equilibrium asset pricing model, Econometrica 5 (1978), no. p. 0
Where the paper says it

Our results rely on fine properties of dynamic mean–variance hedging in general semimartingale markets. 1 Introduction The capital asset pricing model (CAPM) of Treynor [32], Sharpe [29], Lintner [19, 20] and Mossin [24] is one of the first general equilibrium models for financial markets.

46, 1077–1096.

[31] Christophe Stricker, Arbitrage et lois de martingale, Annales de l’I. p. 2

Where the paper says it

, d1}, we have S j t = S j 0 + Mj t + A j t , 0 ≤ t ≤ T, (2. 1) where S j 0 ∈ R and the local martingale part Mj ∈ M2 0,loc are given a priori.

H.P. Probabilit´es et statistiques 26 (1990), 451–460 (fre). p. 0
Where the paper says it

Despite its limitations, it is still one of the cornerstones of modern financial theory and widely used in investment practice; see [18] for a recent overview. Notwithstanding the enormous influence of the CAPM, a rigorous study of existence and uniqueness of CAPM equilibria was initiated only around 1990 by Nielsen [25, 26, 27] and Allingham [1], with more recent important contributions by Berk [2], Dana [11], Hens et al.

[32] Jack Treynor, Toward a theory of market value of risky assets, Unpublished manuscript, 1962. p. 0

Where the paper says it

1 Introduction The capital asset pricing model (CAPM) of Treynor [32], Sharpe [29], Lintner [19, 20] and Mossin [24] is one of the first general equilibrium models for financial markets. Despite its limitations, it is still one of the cornerstones of modern financial theory and widely used in investment practice; see [18] for a recent overview.

[33] Jan Wenzelburger, The two-fund separation theorem revisited, Annals of Finance 6 (2010), 221–239. p. 0

Where the paper says it

, functionals of the form U(µ, σ), where U is quasiconcave, increasing in the mean µ and decreasing in the volatility σ. This is because without distributional assumptions on the returns, the only utility functions that are compatible with the CAPM (more precisely, the two-fund separation theorem) are quadratic utility functions; see the discussion in Berk [2, after Corollary 3.

Auxiliary Results and Proofs

Lemma A.8 establishes the equivalence between the zero solution to the MVH problem and a local P-martingale condition for S j. p. 7

Where the paper says it

Let (1, S) be a local L 2 -market with η k ∈ Θ(S). Suppose that there is a unique solution ϑ MVH(1) to the pure investment problem (3.

Proposition A.7 shows that if Zt 6= 0 P-a.s., then (1, S) satisfies uniqueness of value processes. p. 2
Where the paper says it

We assume that the price processes S (1) and S (2) are not given a priori but rather determined in equilibrium between K agents trading in the market. We assume that the initial value and volatility structure of the financial assets are predetermined and known by the market participants, i.

The proof of Theorem 3. p. 1
Where the paper says it

We consider general semimartingale markets and assume that the agents receive endowments at the terminal time T that are partly unhedgeable. In the case of of quadratic utility, our proof of existence and uniqueness of an equilibrium is based on the construction of a nonstandard type of representative agent, i.

7 involves showing that the candidate equilibrium price process S j is a local L2-semimartingale, which ensures condition 3) of Definition 2. p. 2
Where the paper says it

2) the rest of the price process (S j )0≤t<T is to be determined by the market in equilibrium. Finally, we assume that each asset S j is a local L 2 -semimartingale for j ∈ {1, .

6. The proof of Lemma 4. p. 1
Where the paper says it

We consider general semimartingale markets and assume that the agents receive endowments at the terminal time T that are partly unhedgeable. In the case of of quadratic utility, our proof of existence and uniqueness of an equilibrium is based on the construction of a nonstandard type of representative agent, i.

9 relies on Theorem 3. p. 0
Where the paper says it

, functionals of the form U(µ, σ), where U is quasiconcave, increasing in the mean µ and decreasing in the volatility σ. This is because without distributional assumptions on the returns, the only utility functions that are compatible with the CAPM (more precisely, the two-fund separation theorem) are quadratic utility functions; see the discussion in Berk [2, after Corollary 3.

7 to establish that S(¯γ) satisfies Assumption 4. p. 2
Where the paper says it

1 Financial market We work on a filtered probability space (Ω, F, F = (Ft)0≤t≤T , P) with a fixed finite time horizon T ∈ (0,∞). We assume that the filtration F satisfies the usual conditions of right-continuity and completeness.

3. The proof of Proposition 4. p. 1
Where the paper says it

We consider general semimartingale markets and assume that the agents receive endowments at the terminal time T that are partly unhedgeable. In the case of of quadratic utility, our proof of existence and uniqueness of an equilibrium is based on the construction of a nonstandard type of representative agent, i.

12 involves solving the fixed-point equation γ̄ = ˜γ(¯γ) derived from the dynamic theory. p. 0
Where the paper says it

We then use this link to study a fixed-point problem that establishes existence (and uniqueness in a suitable class) of linear mean–variance equilibria. Our results rely on fine properties of dynamic mean–variance hedging in general semimartingale markets.

The proof of Theorem 4. p. 1
Where the paper says it

We consider general semimartingale markets and assume that the agents receive endowments at the terminal time T that are partly unhedgeable. In the case of of quadratic utility, our proof of existence and uniqueness of an equilibrium is based on the construction of a nonstandard type of representative agent, i.

14 concludes that a mean–variance equilibrium exists if and only if the aggregate risk tolerance PK k=1 λ k is larger than the uncertainty of the aggregate endowment Ξ, as measured by ess sup Ξ¯ − EP [Ξ] ¯ ≥ 0. p. 1
Where the paper says it

Because “mean–variance efficient portfolios approximately maximise expected utility for a wide range of risk-averse (concave) utility functions” (see Markowitz [21, 22]), our results may also provide approximations to equilibrium prices for other utility functions. In most of the extant literature on the existence and uniqueness of CAPM equilibria, it is assumed that the aggregate random endowment is tradable so that the financial market is complete.

The paper concludes that equilibria with aggregate risk tolerance ¯γ > γ¯0 are the most meaningful from an economic point of view. p. 5
Where the paper says it

8) where U Q k (x) := 2γkx − x 2 for some risk-tolerance γk ∈ R. The parameter γk is also sometimes called the bliss point of agent k, as it is the optimal wealth that the agent would like to attain at time T in order to maximise their utility.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Enhanced handling of general semimartingale market dynamics by allowing equilibrium prices driven by general semimartingales in both discrete and continuous time settings. This allows AI systems to model more complex financial environments beyond standard Brownian motion assumptions.

  2. Rigorous proof structure for existence and uniqueness of quadratic mean-variance equilibria in incomplete markets, based on the construction of a nonstandard type of representative agent. This capability enables the AI to determine if a market structure is viable under real-world frictions like unhedgeable endowments.

  3. Explicit characterization of the mean-variance efficient frontier using the triplet (l, ck, epsilon2k), which allows AI to precisely map risk tolerance parameters to optimal trading strategies in linear mean-variance preferences. This improves the system's ability to generate optimal portfolio allocations for linear utility functions.

  4. Derivation of an explicit solution for the unique optimal strategy in linear mean-variance preferences as ϑk (lambda k over l), which provides a direct computational path for finding equilibrium strategies under specific market assumptions. This accelerates the AI's task in solving complex, non-quadratic utility optimization problems.

  5. Development of necessary and sufficient conditions for the existence of quadratic equilibrium markets based on martingale measures, including conditions like Zbar t 6= 0 and Zbar t-1 = 0 for all t P-a.s. This allows the AI to verify market stability and pricing feasibility under specific structural constraints.

  6. Ability to distinguish between mean-variance equilibria that satisfy Assumption 4.3 (ELMM exists) and those that do not, providing a nuanced understanding of market completeness and arbitrage opportunities in different settings. This improves risk assessment capabilities for complex asset classes like real assets.

Abstract

We study continuous-time CAPM equilibria with general mean-variance preferences of the form U(μ,σ), where U is strictly quasiconcave, increasing in the mean μ and decreasing in the volatility σ of terminal wealth. Our framework accommodates both financial and productive assets with nontradeable endowments. We allow for equilibria driven by general semimartingales in the case of linear mean-variance preferences and by continuous semimartingales in the case of general mean-variance preferences. The key insight is a connection between mean-variance efficiency and quadratic utility maximisation. We first characterise quadratic equilibria through an explicit family of equilibrium price processes indexed by the aggregate risk tolerance. We then show that each agent's optimal mean-variance strategy corresponds to a quadratic optimum with an endogenously determined risk tolerance. This correspondence reduces the mean-variance equilibrium problem to a one-dimensional fixed-point problem, whose solvability we establish using fine properties of (dynamic) mean-variance hedging and a novel stability result for quadratic BSDEs with BMO growth at the origin. Finally, examples illustrate how unspanned stochastic-volatility risk interacts with aggregate endowments to affect equilibrium asset prices, and how nonlinear mean-variance preferences influence equilibrium risk tolerance and asset-price comparative statics.

Sources

Related papers