Abstracts

Efficient prediction for nonlinear autoregressive models

Wolfgang Wefelmeyer (University of Siegen)

Conditional expectations given past observations in stationary time series are usually estimated directly by kernel estimators. Such estimators do not converge at the (parametric) root n rate. For nonlinear autoregressive models, we construct estimators of conditional expectations that converge at the root n rate and are asymptotically efficient in the sense of Hajek and Le Cam. Our estimators exploit the independence of the innovations; they are smoothed and weighted von Mises statistics based on the residuals. The estimators are studied using two results of independent interest: convergence rates for weitalksghted residual-based innovation density estimators, and uniform stochasic expansions of residual-based von Mises processes. This is joint work with Ursula U. Mueller (Bremen) and Anton Schick (Binghamton).

Semiparametric Time Series Models with Non-IID Errors

Feike C. Drost and Bas J.M. Werker (Tilburg University)

We study estimation of parametric components in general semiparametric time series models. The time series models are not assumed to be driven by a sequence of independent innovations with unknown distribution as is the case in a more traditional semiparametric time series approach. We assume just that the innovations are stationary and the dependence between the innovations is seen as a nonparametric nuisance parameter in addition to the marginal distribution of the innovations. We obtain a LAN result under quite natural and economical conditions implying a lower bound on the asymptotic performance of efficient estimators. We also study the influence of additional knowledge on the conditional innovation distribution and discuss the lower bounds obtained for several cases.

Mean-variance hedging in large financial markets

Luciano Campi

We consider a mean-variance hedging problem for an arbitrage-free large financial market, i.e. a financial market with countably many risky assets modelled by a sequence of continuous semimartingales. By using the stochastic integration theory for cylindrical martingales developed in Mikulevicius and Rozovskii (1998), we extend the results about a change of numeraire and mean-variance hedging of Gourieroux, Laurent and Pham (1998) to this setting. We also consider, for all n≥1, the market formed by the first n risky assets and study the solutions to the n-dimensional mean-variance hedging problem associated and their behaviour when n tends to infinity.

Integration and arbitrage in fractional Black-Scholes model

Tommi Sottinen

It has been proposed that the arbitrage possibility in the fractional Black-Scholes model depends on the definition of the stochastic integral. More precisely, if one uses the Wick-Ito-Skorohod integral one obtains an arbitrage-free model. However, this integral does not allow economical interpretation. On the other hand it is easy to give abrbitrage examples in continuous time trading with self-financing strategies, if one uses the Riemann-Stieltjes integral.
We study the connection between two different notions of self-financing portfolios in the fractional Black-Scholes model by applying the known connection between these two integrals. In particular, we give an economical interpretation of the proposed arbitrage-free model in terms of Riemann-Stieltjes integrals.

This is joint work with Esko Valkeila.