Path-conditional forward volatility

In derivatives modelling, it has often been necessary to make assumptions about the volatility of the underlying variable over the life of the contract. This can involve specifying an exact trajectory, as in the Black and Scholes (1973), Merton (1973) or Black (1976) models; one that depends on the level of the underlying variable as in the local volatility models of Dupire (1994), Derman and Kani (1994) and Rubinstein (1994); or fixing the parameters of a more general stochastic volatility process as in Hull and White (1987) or Heston (1993). These forward-looking assumptions are by their very nature destined to be disproved, and what is more are at odds with the frequent model recalibration that (rightly) takes place in practice. In Carey (2005), the Black-Scholes analytical framework is extended, via the definition of higher-order volatilities and the derivation of moment formulae for the case where they are deterministic. In this paper, we show that the same formulae can be obtained under markedly weaker assumptions, which leave the future volatilities unspecified. Instead, we impose constraints on new, related quantities, which we term "path-conditional forward volatilities." Under this scheme, the model inputs are no longer the future spot volatilities, but rather their forward counterparts. One consequence, we show, is that contrary to conventional wisdom, the Black-Scholes formula can in principle be used without any reference to future volatility.