No arbitrage of the first kind and local martingale numéraires (Q331366)

From MaRDI portal
scientific article
Language Label Description Also known as
English
No arbitrage of the first kind and local martingale numéraires
scientific article

    Statements

    No arbitrage of the first kind and local martingale numéraires (English)
    0 references
    0 references
    0 references
    0 references
    27 October 2016
    0 references
    In recent years, there has been a substantial interest in the characterization and the properties of the \textit{no arbitrage of the first kind} (NA\(_1\)) condition. In particular, NA\(_1\) is weaker than the classical NFLVR condition and it has been shown that NA\(_1\) is the minimal no-arbitrage-type condition under which portfolio optimization problems can be meaningfully solved, see for instance [\textit{I. Karatzas} and \textit{C. Kardaras}, Finance Stoch. 11, No. 4, 447--493 (2007; Zbl 1144.91019)] and [\textit{T. Choulli} et al., Finance Stoch. 19, No. 4, 719--741 (2015; Zbl 1358.91091)]. The present paper contains a novel characterization of NA\(_1\) in the setting of a general multidimensional semimartingale financial model, complementing earlier studies on this condition. More specifically, it is well-known (see e.g. [Karatzas and Kardaras, loc. cit.]) that NA\(_1\) is equivalent to the existence of a \textit{supermartingale numéraire}, i.e., a strictly positive portfolio such that the value process of every non-negative portfolio becomes a supermartingale when denominated in units of the numéraire portfolio. In this paper, this property is refined and it is shown that, in an arbitrary neighborhood (in the total variation distance) of the original probability measure \(P\), one can find an equivalent probability measure \(P'\) such that the supermartingale numéraire becomes a \textit{local martingale numéraire}, i.e., the value process of every non-negative portfolio becomes a local martingale under \(P'\) when denominated in units of the numéraire portfolio. The proof is based on a careful study of ratios of stochastic exponentials together with an application of fundamental results from no-arbitrage theory.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    numéraire
    0 references
    arbitrage
    0 references
    local martingale
    0 references
    separating measure
    0 references
    arbitrage of first kind
    0 references
    fundamental theorem of asset pricing
    0 references
    0 references