Asymptotic replication with modified volatility under small transaction costs (Q287666)

The article deals with the problem of hedging derivative instruments in local volatility models with transaction costs. The authors assume that the asset price, \(S_t\), follows the stochastic differential equation \(dS_t=\Theta_t dt + \sigma(S_t,t)dB_t\), where \(B\) is a Wiener process. They consider some benchmark strategy \(q\), given by the value process \(\Pi^q =q(S_t, t)\) and the portfolio \(X^q_t=\partial_S q(S_t, t)\). Then the problem of finding a strategy \(X\) that approximates the benchmark strategy \(X^q\) in the presence of transaction costs is formulated. It is assumed that the approximated strategy \(X\) should lie in some corridor, given by a function \(b(S_t,t)\), around the benchmark strategy. The transaction costs are given by \(\kappa \int_0^t \lambda(S_u, u)d\|X\|_u\), where \(\kappa\) is a proportionality coefficient (assumed to be small). The authors consider the tracing error -- the difference between the value of the benchmark strategy and the approximated strategy, \(\varepsilon_t = \Pi^q_t - \Pi_t\) -- and its asymptotic properties. The main results of the article are presented in two theorems. The first one (Theorem 3.1) states that when transaction costs tend to zero (\(\kappa \to 0\)) the tracking error (normalized by \(\kappa^{-1}\)) is asymptotically Gaussian. Then, the authors consider the problem of minimizing the asymptotical variance of the tracking error. In the second main theorem (Theorem 3.7), they provide a lower bound on this variance. Furthermore, they provide a construction of a sequence of trading strategies which are asymptotically optimal. In the rest of the paper, the authors provide results of numerical simulations for hedging a call option in the Black-Scholes model with proportional transaction costs and compare them with \textit{H. Leland}'s strategy [``Optimal pricing and replication with transaction costs'', J. Finance 40, No. 5, 1283--1301 (1985; \url{doi:10.1111/j.1540-6261.1985.tb02383.x})]. Section 4 contains a heuristic derivation of Theorem 3.1, based on a discretization of the model. The authors also use this approach to compare their approach with those of \textit{M. Denis} and \textit{Y. Kabanov} [Finance Stoch. 14, 625--667 (2010; Zbl 1233.91262)], \textit{M. Fukasawa} [``Conservative delta hedging under transaction costs'', in: Recent advances in financial engineering. Singapore: World Scientific. 55--72 (2012)] and \textit{A. E. Whalley} and \textit{P. Wilmott} [Math. Finance 7, No. 3, 307--324 (1997; Zbl 0885.90019)].