FTAP in finite discrete time with transaction costs by utility maximization (Q2255008)

The authors prove the Fundamental Theorem of Asset Pricing in finite discrete time with proportional transaction costs by utility maximization. Several proofs of this result are available, the main result of the paper is actually a theorem from [\textit{W. Schachermayer}, Math. Finance 14, No. 1, 19--48 (2004; Zbl 1119.91046)], but the authors argue that they deliver the first ``proof based on the economically meaningful and intuitive approach via utility maximization''. Let \((\Omega,{\mathcal F},({\mathcal F}_{t})_{t=0}^{T},{\mathbb P})\) be a filtered probability space. Let \(({\underline S}_{t},\overline S_{t})_{t=0}^{T}\) be a two-dimensional \(({\mathcal F}_{t})_{t=0}^{T}\)-adapted process with \(0 < {\underline S}_{t} \leq \overline S_{t}\) for every \(t=0,\dots,T\), where \({\underline S}_{t}\) and \(\overline S_{t}\) stand for the bid and ask prices. The \textit{solvency cone} of \(({\underline S}_{t},\overline S_{t})\) is \[ K({\underline S}_{t},\overline S_{t}) = \{(v_{0},v_{1}) : v_{0} + v_{1} {\underline S}_{t} \geq 0,~v_{0} + v_{1} {\overline S}_{t} \geq 0\}. \] \(({\underline S}_{t},\overline S_{t})_{t=0}^{T}\) is said to satisfy the \textit{no-arbitrage condition} if the only self-financing portfolio which lies almost surely in the cones of portfolios available at price zero \(-K({\underline S}_{t},\overline S_{t})\) \((t=0,\dots,T)\) is a portfolio which is exchangeable to zero, i.e.\ lies in \(K({\underline S}_{t},\overline S_{t}) \cap -K({\underline S}_{t},\overline S_{t})\) \((t=0,\dots,T)\), almost surely. \(({\underline S}_{t},\overline S_{t})_{t=0}^{T}\) is said to satisfy the \textit{robust no-arbitrage condition} if there exists a process \(({\underline R}_{t},\overline R_{t})_{t=0}^{T}\) satisfying the no-arbitrage condition such that \([{\underline R}_{t},\overline R_{t}] \subseteq [{\underline S}_{t},\overline S_{t}]\) \((t=0,\dots,T)\), moreover \([{\underline R}_{t},\overline R_{t}] \subset ({\underline S}_{t},\overline S_{t})\) for every \(t\) with \({\underline S}_{t} \neq \overline S_{t}\). The authors prove the following theorem: \(({\underline S}_{t},\overline S_{t})_{t=0}^{T}\) satisfies the robust no-arbitrage condition if and only if there exists an equivalent probability measure \(\mathbb Q\) and a \(\mathbb Q\)-martingale \((S_{t})_{t=0}^{T}\) such that \(S_{t} \in [{\underline S}_{t},\overline S_{t}]\) \((t=0,\dots,T)\) and \(S_{t} \in ({\underline S}_{t},\overline S_{t})\) for every \(t\) with \({\underline S}_{t} \neq \overline S_{t}\). To prove this result, the authors take a utility function and consider optimal self-financing portfolios according to the bid-ask prices \(({\underline S}_{t},\overline S_{t})_{t=0}^{T}\). After establishing that the optimal terminal wealth can be achieved, the Radon-Nikodym derivative of \({\mathbb Q}\) with respect to \({\mathbb P}\) is expressed using the utility of the optimal terminal wealth, and the consistent price system \((S_{t})_{t=0}^{T}\) is obtained by induction on \(t\).