Towards a generalization of Dupire's equation for several assets (Q1018345)

\textit{H. Dupire} [Risk 7, 18--20 (1994)] considered a model for the dynamics of an underlying asset in which the volatility depends on time \(t\) and on the stock price \(S\). More precisely, \[ \frac{dS}{S}=\mu dt+\sigma(t,S)dw.\tag{*} \] (Here, \(W\) is a Brownian motion, \(\mu\) is a drift coefficient, and \(\sigma(\cdot, \cdot)\) is the volatility of the underlying asset.) Dupire has shown that in the model (*) the volatility can, in principle, be recovered from market data if the price of European options on the underlying asset were known for all strike prices \(K\) and all maturity dates \(T\). In fact, Dupire has shown that \[ \sigma= \left(\left[\frac{\partial C}{\partial T}-(r(t)-D(t))\left(C-K\frac{\partial C} {\partial T}\right)\right]\left/\left(\frac 12 K^2\frac{\partial^2C}{\partial K^2}\right)\right.\right)^{1/2}.\tag{**} \] Here, \(C(t,S_t,T,K)\) is the undiscounted European call option price, \(r(t)\) is the risk-free interest rate, and \(D(t)\) is the dividend rate. ((**) holds under the usual assumptions of liquidity, absence of arbitrage, and transaction costs.) The authors obtain a generalization of Dupire 's equation for the multi-asset context under additional assumptions (the most restrictive being that the volatility does not depend on asset prices).