Abstract
We study the response of the optimal investment problem to small changes of the stock price dynamics. Starting with a multidimensional semimartingale setting of an incomplete market, we suppose that the perturbation process is also a general semimartingale. We obtain second-order expansions of the value functions, first-order corrections to the optimisers, and provide the adjustments to the optimal control that match the objective function up to the second order. We also give a characterisation in terms of the risk-tolerance wealth process, if it exists, by reducing the problem to the Kunita–Watanabe decomposition under a change of measure and numéraire. Finally, we illustrate the results by examples of base models that allow closed-form solutions, but where this structure is lost under perturbations of the model where our results allow an approximate solution.
Similar content being viewed by others
Notes
We should like to thank a referee for pointing out these ideas.
References
Anthropelos, M., Robertson, S., Spiliopoulos, K.: The pricing of contingent claims and optimal positions in asymptotically complete markets. Ann. Appl. Probab. 27, 1778–1830 (2017)
Backhoff Veraguas, J., Silva, F.: Sensitivity analysis of expected utility maximization in incomplete Brownian market models. Math. Financ. Econ. 12, 387–411 (2018)
Bank, P., Körber, L.: Merton’s optimal investment problem with jump signals. SIAM J. Financ. Math. 13, 1302–1325 (2022)
Barrieu, P., El Karoui, N.: Monotone stability of quadratic semimartingales with applications to unbounded general quadratic BSDEs. Ann. Probab. 41, 1831–1863 (2013)
Černý, A., Kallsen, J.: On the structure of general mean–variance hedging strategies. Ann. Probab. 35, 1479–1531 (2007)
Czichowsky, C., Schweizer, M.: Cone-constrained continuous-time Markowitz problems. Ann. Appl. Probab. 23, 764–810 (2013)
Delbaen, F., Schachermayer, W.: A general version of the fundamental theorem of asset pricing. Math. Ann. 300, 463–520 (1994)
Delbaen, F., Schachermayer, W.: The Mathematics of Arbitrage. Springer, Berlin (2006)
Föllmer, H., Schweizer, M.: The minimal martingale measure. In: Cont, R. (ed.) Encyclopedia of Quantitative Finance, pp. 1200–1204. Wiley, New York (2010)
Goll, T., Kallsen, J.: A complete explicit solution to the log-optimal portfolio problem. Ann. Appl. Probab. 13, 774–799 (2003)
Golub, G.H., Van Loan, C.F.: Matrix Computations, 3rd edn. Johns Hopkins University Press, Baltimore (1996)
Gouriéroux, C., Laurent, J., Pham, H.: Mean–variance hedging and numéraire. Math. Finance 8, 179–200 (1998)
Guasoni, P., Robertson, S.: Static fund separation of long-term investments. Math. Finance 25, 789–826 (2015)
Henderson, V.: Valuation of claims on nontraded assets using utility maximization. Math. Finance 12, 351–373 (2002)
Henderson, V., Hobson, D.: Real options with constant relative risk aversion. J. Econ. Dyn. Control 27, 329–355 (2002)
Herrmann, S., Muhle-Karbe, J., Seifried, F.T.: Hedging with small uncertainty aversion. Finance Stoch. 21, 1–64 (2017)
Horst, U., Hu, Y., Imkeller, P., Réveillac, A., Zhang, J.: Forward–backward systems for expected utility maximization. Stoch. Process. Appl. 124, 1813–1848 (2014)
Hu, Y., Imkeller, P., Müller, M.: Utility maximization in incomplete markets. Ann. Appl. Probab. 15, 1691–1712 (2005)
Jacod, J., Shiryaev, A.N.: Limit Theorems for Stochastic Processes, 2nd edn. Springer, Berlin (2003)
Jeanblanc, M., Klöppel, S., Miyahara, Y.: Minimal \(f^{q}\)-martingale measure for exponential Lévy processes. Ann. Appl. Probab. 17, 1615–1638 (2007)
Jeanblanc, M., Mania, M., Santacroce, M., Schweizer, M.: Mean–variance hedging via stochastic control and BSDEs for general semimartingales. Ann. Appl. Probab. 22, 2388–2428 (2012)
Kallsen, J.: Optimal portfolios for exponential Lévy processes. Math. Methods Oper. Res. 51, 357–374 (2000)
Kallsen, J.: Derivative pricing based on local utility maximization. Finance Stoch. 6, 115–140 (2002)
Kallsen, J., Muhle-Karbe, J.: Utility maximization in models with conditionally independent increments. Ann. Appl. Probab. 20, 2162–2177 (2010)
Kallsen, J., Muhle-Karbe, J.: The general structure of optimal investment and consumption with small transaction costs. Math. Finance 27, 659–703 (2017)
Karatzas, I., Kardaras, C.: The numéraire portfolio in semimartingale financial models. Finance Stoch. 11, 447–493 (2007)
Karatzas, I., Kardaras, C.: Portfolio Theory and Arbitrage: A Course in Mathematical Finance. Am. Math. Soc., Providence (2021)
Kramkov, D., Schachermayer, W.: The asymptotic elasticity of utility functions and optimal investment in incomplete markets. Ann. Appl. Probab. 9, 904–950 (1999)
Kramkov, D., Schachermayer, W.: Necessary and sufficient conditions in the problem of optimal investment in incomplete markets. Ann. Appl. Probab. 13, 1504–1516 (2003)
Kramkov, D., Sîrbu, M.: On the two-times differentiability of the value functions in the problem of optimal investment in incomplete markets. Ann. Appl. Probab. 16, 1352–1384 (2006)
Kramkov, D., Sîrbu, M.: Sensitivity analysis of utility-based prices and risk-tolerance wealth process. Ann. Appl. Probab. 16, 2140–2194 (2006)
Laurent, J.P., Pham, H.: Dynamic programming and mean–variance hedging. Finance Stoch. 3, 83–110 (1999)
Liu, J.: Portfolio selection in stochastic environments. Rev. Financ. Stud. 20, 1–39 (2007)
Monoyios, M.: Malliavin calculus method for asymptotic expansion of dual control problems. SIAM J. Financ. Math. 4, 884–915 (2013)
Mostovyi, O.: Necessary and sufficient conditions in the problem of optimal investment with intermediate consumption. Finance Stoch. 19, 135–159 (2015)
Mostovyi, O.: Asymptotic analysis of the expected utility maximization problem with respect to perturbations of the numéraire. Stoch. Process. Appl. 130, 4444–4469 (2020)
Mostovyi, O., Sîrbu, M.: Sensitivity analysis of the utility maximization problem with respect to model perturbations. Finance Stoch. 23, 595–640 (2019)
Pham, H., Rheinländer, T., Schweizer, M.: Mean–variance hedging for continuous processes: new proofs and examples. Finance Stoch. 2, 173–198 (1998)
Protter, P.: A comparison of stochastic integrals. Ann. Probab. 7, 276–289 (1979)
Protter, P.: Stochastic Integration and Differential Equations. Springer, Berlin (2004)
Robertson, J., Rosenberg, M.: The decomposition of matrix-valued measures. Mich. Math. J. 15, 353–368 (1968)
Robertson, S.: Pricing for large positions in contingent claims. Math. Finance 27, 746–778 (2017)
Santacroce, M., Trivellato, B.: Forward backward semimartingale systems for utility maximization. SIAM J. Control Optim. 52, 3517–3537 (2014)
Schweizer, M.: Martingale densities for general asset prices. J. Math. Econ. 21, 363–378 (1992)
Schweizer, M.: On the minimal martingale measure and the Föllmer–Schweizer decomposition. Stoch. Anal. Appl. 13, 573–599 (1995)
Zariphopoulou, T.: A solution approach to valuation with unhedgeable risks. Finance Stoch. 5, 61–82 (2001)
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Competing Interests
The authors declare no competing interests.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Oleksii Mostovyi has been supported by the National Science Foundation under grant No. DMS-1848339 (2019-2024). Mihai Sîrbu has been supported by the National Science Foundation under grant No. DMS-1908903 (2019-2023). Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. The authors would like to thank Martin Schweizer, whose questions, comments and suggestions led to a substantial improvement of the paper.
Appendix: The structure of \(\mathcal{M}^{\infty}\) and \(\mathcal{N}^{\infty}\)
Appendix: The structure of \(\mathcal{M}^{\infty}\) and \(\mathcal{N}^{\infty}\)
We recall that Mostovyi [36, Lemma 4.1] shows that every element of \(\mathcal{M}^{\infty}\) can be represented as a stochastic integral with respect to \(R^{\pi}\). The following lemma establishes the opposite direction.
Lemma A.1
Fix \(x>0\) and set \(y = u_{x}(x,0)\). Suppose \(M\in \mathcal{H}^{2}_{ \mathrm{loc}}(\mathbb{P})\), that (2.2) and Assumptions 2.1and 2.2hold and that \(R^{\pi}\) is sigma-bounded. Then we have
Remark A.2
The proof goes through without the sigma-boundedness assumption. The latter is imposed to ensure that \(\mathcal{M}^{\infty}\) is non-degenerate and that the closure of \(\mathcal{M}^{\infty}\) in \(\mathcal{H}^{2}_{0}(\mathbb{R})\) is equal to \(\mathcal{M}^{2}\). Also, the proof goes through with NUPBR or, equivalently, (2.10), instead of \(M\in \mathcal{H}^{2}_{\mathrm{loc}}(\mathbb{P})\) and (2.2), and with the Inada conditions instead of Assumption 2.1; all we need is that together with Assumption 2.2, the standard assertions of utility maximisation theory hold.
Proof of Lemma A.1
Let \(\alpha \) be predictable and \(R^{\pi}\)-integrable and such that \(\alpha \cdot R^{\pi}\) is bounded. Then there exists a constant \(C>0\) such that \(C + \alpha \cdot R^{\pi}\) is strictly positive. By Jacod and Shiryaev [19, Theorem II.8.3], there exists a predictable \(R^{\pi}\)-integrable process \(\widetilde{\alpha}\) such that
where the second equality uses (3.2). We deduce that the bounded process \(\alpha \cdot R^{\pi}\) admits the representation
which is an element of \(\mathcal{M}^{\infty}\) by the definition of \(\mathcal{M}^{\infty}\). As \(\alpha \) was arbitrary, the proof is complete. □
Lemma A.3
Fix \(x>0\) and set \(y = u_{x}(x,0)\). Impose the assumptions of Lemma A.1and that both \(M\) and \(H\) are in \(\mathcal{H}^{2}_{\mathrm{loc}}(\mathbb{P})\). Recall that H is defined in Assumption 2.3and is such that \(Y = y\mathcal{E}(H)\). Then we have
Proof
Take \(N\in \mathcal{H}^{2}_{\mathrm{loc}}(\mathbb{P})\) such that \(N^{H}\) is bounded and fix \(\widetilde{M}\in \mathcal{M}^{\infty}\). By [36, Lemma 4.1], we have \(\widetilde{M} = \alpha \cdot R^{\pi}\) for some predictable \(R^{\pi}\)-integrable process \(\alpha \). Let us approximate \(\alpha \) by the \(M\)-integrable processes
where every component of \(\alpha \) is truncated from above by \(n\) and below by \(-n\) and where \(\tau _{n}\), \(n\in \mathbb{N}\), is a localising sequence for both \(M\) and \(N\). Then for a fixed \(n\in \mathbb{N}\) and every stopping time \(\tau \), similarly to Lemma 5.13, we get
As \(N^{H}\) is bounded and \(\alpha ^{n}\cdot R^{\pi}\), \(n\in \mathbb{N}\), converges to \(\alpha \cdot R^{\pi}\) in \(\mathcal{H}^{2}(\mathbb{R})\), we deduce from (A.1) that \(N^{H}\) is orthogonal to \(\alpha \cdot R^{\pi}\). Now from Lemma A.1, we deduce that \(N^{H}\) is orthogonal to \(\mathcal{M}^{\infty}\). Since additionally, the closure of \(\mathcal{M}^{\infty}\) in \(\mathcal{H}^{2}_{0}(\mathbb{R})\) is equal to \(\mathcal{M}^{2}\) by Kramkov and Sîrbu [30, Lemma 6], we get
To show the opposite inclusion, we proceed as follows. We fix \(K\in \mathcal{N}^{\infty}\) and set \(N := K+ [K, H] = K^{-H}\). Then \(N\) is locally square-integrable under ℙ because \(K\) is bounded and \(H\) is locally ℙ-square-integrable. We suppose that \(\mathcal{E}(K)>0\), as otherwise we may multiply \(K\) by a sufficiently small constant \(\varepsilon \) and conduct the proof for \(\varepsilon K\).
For \(\alpha = -\pi \), as \(\mathcal{E}(\alpha \cdot R^{\pi}) = \frac {1}{\mathcal{E}(\pi \cdot R)}>0\), using the sigma-boundedness of \(R^{\pi}\) and the Ansel–Stricker theorem (see [8, Corollary 7.3.8]), we deduce that \(\mathcal{E}(\alpha \cdot R^{\pi})\) is a local martingale under ℝ and hence in \(\mathcal{H}^{1}_{\mathrm{loc}}(\mathbb{R})\). Let \(\sigma _{n}\), \(n\in \mathbb{N}\), be a localising sequence for \(\mathcal{E}(\alpha \cdot R^{\pi})\) such that on \([\!\![0,\sigma _{n}]\!\!]\), \(\mathcal{E}(\alpha \cdot R^{\pi})\) is in \(\mathcal{H}^{1}(\mathbb{R})\), where by boundedness of \(K\), we also suppose that \(\mathcal{E}(K)\) is bounded on \([\!\![0,\sigma _{n}]\!\!]\). Further, as \(R^{\pi}\) is sigma-bounded, we can use [30, Theorem 4] to approximate \(\mathcal{E}(\alpha \cdot R^{\pi})\) on \([\!\![0,\sigma _{n}]\!\!]\) in \(\mathcal{H}^{1}(\mathbb{R})\) by some bounded stochastic integrals with respect to \(R^{\pi}\). These integrals are in \(\mathcal{M}^{\infty}\) by Lemma A.1. By convergence in \(\mathcal{H}^{1}(\mathbb{R})\) and boundedness of \(\mathcal{E}(K)\) on \([\!\![0,\sigma _{n} ]\!\!]\), we obtain
and by a change of measure, we have
Comparing (A.2) and (A.3), we deduce that \(\mathcal{E}\left (N+H\right )\) is a local martingale under ℙ. As \(\mathcal{E}\left (N+H\right )\) and \(\mathcal{E}\left (N+H\right )_{-}\) are both non-vanishing by construction because \(\mathcal{E}\left (N+H\right ) = \mathcal{E}(K)\mathcal{E}(H)\), the stochastic logarithm of \(\mathcal{E}\left (N+H\right )\) is well defined by [19, Theorem II.8.3] and is equal to \(N+H\) by [19, Corollary II.8.7]. Further, from [19, Theorem II.8.3] and Protter [40, Theorem III.29], we conclude that \(N+H\) is a local martingale under ℙ. Moreover, \(N\) is locally ℙ-square-integrable as seen in the preceding paragraph and \(H\in \mathcal{H}^{2}_{\mathrm{loc}}(\mathbb{P})\) by assumption, and so we conclude that \(N\in \mathcal{H}^{2}_{\mathrm{loc}}(\mathbb{P})\).
Second, we show that \(N\) is orthogonal to \(M\). For this, take an element of \(\mathcal{M}^{\infty}\) of the form \(\alpha \cdot R^{\pi}\), where by sigma-boundedness, we suppose that each component of \(\alpha \) takes values in \((0,1]\). Choose a stopping time \(\sigma \) such that \(\mathbb{E}[\int _{0}^{\sigma }\alpha ^{\top}_{s} d\left \langle M \right \rangle _{s} \alpha _{s} ]<\infty \) and \(\mathbb{E}[\left \langle N \right \rangle _{\sigma}]<\infty \). Then as \(K = N^{H}\in \mathcal{N}^{\infty}\), similarly to the proof of Lemma 5.13, we have
From (A.4), we deduce that \(\left \langle \alpha \cdot M, N \right \rangle \) is an ℝ-martingale on \([\!\![0, \sigma ]\!\!]\). Further, since \(\left \langle \alpha \cdot M, N \right \rangle \) is predictable and of finite variation, we deduce that \(\left \langle \alpha \cdot M, N \right \rangle \equiv 0\) on \([\!\![0, \sigma ]\!\!]\). As \(M\) and \(N\) are locally square-integrable and each component of \(\alpha \) in the previous paragraph was \((0,1]\)-valued, thus non-vanishing, we can deduce by localisation that each component of \(M\) is orthogonal to \(N\) on \([0,T]\). □
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Mostovyi, O., Sîrbu, M. Quadratic expansions in optimal investment with respect to perturbations of the semimartingale model. Finance Stoch 28, 553–613 (2024). https://doi.org/10.1007/s00780-024-00532-6
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00780-024-00532-6