Price operators analysis in \(L_p\)-spaces (Q2492715)

The mathematical investigations in this paper regarding monotone linear operators in \(L_{p}\)-spaces arise in connection with a model of asset pricing as a continuous time stochastic process over a finite horizon \([ 0,T] \) with increasing \(\sigma\)-algebras \(\mathcal{U}_{t}\) representing the information of the market events up to time. Let \( ( \Omega ,\mathcal{U},\mathcal{P}) \) be a probability space where \(\mathcal{U=U}_{T}\), and \(\mathcal{P}:\mathcal{U}\to [ 0,1 ] \) denotes the underlying probability measure. There is a riskless security with deterministic return \(R_{t}>0\) at time \(t\). There are some risky securities, for any of which the price \(X_{t}\) at time \(t\) is a real random variable measurable with respect to \(\mathcal{U}_{t}\). A probability measure \(\mathcal{P}^{0},\) equivalent to \(\mathcal{P},\) is risk-neutral if the expected return from all of the risky securities, under \(\mathcal{P}^{0},\) is the same as that from the riskless security, i.e., \[ E^{0}\left[\frac{X_{t}-X_{s}}{X_{s}}\,\biggl|\, \mathcal{U}_{s} \right]=\frac{R_{t}-R_{s}}{R_{s} }, \quad 0\leq s<t\leq T. \] Since \(R_{t}^{-1}\) is the present or discounted value of \(1\) unit of the riskless security (money) received at time \(t\), in units of the riskless security (money) received at time \(0,R_{t}^{-1}X_{t}\) is the discounted price process for the risky security. More generally, if \(s\) and \(t\) are two points of time with \(s<t\) writing the security prices \(X_{t}\) as \(X\) (the positive real \(\mathcal{U}_{t}\)-measurable random variables, \(X:=X( \omega ) \geq 0\), \(\omega \in \Omega ,\) constituting the positive convex cone \(L^{+})\) and the relationship between prices at \(s\) and those at \(t\) implied by risk neutrality as \(x( X) ,\) \(x( X) :=X_{s}\), \(X\in L^{+},\) is a linear operator of the form \( x( X) =E^{0}[ R^{-1}X\mid B]\), \(X\in L^{+},\) where \( R^{-1}:=\frac{R_{s}}{R_{t}}>0\) is the discount factor between \(s\) and \(t\) and the \(\sigma -\)algebra \(B:=\mathcal{U}_{s}.\) This price operator is considered in the standard \(L_{p}-\)space, \(L_{p}:=L_{p}( \Omega , \mathcal{U},\mathcal{P}) ,1\leq p<\infty ,\) of the random variables \(X\) with the norm \(\|X\|:=( E|X|^{p}) ^{\frac{ 1}{p}}.\) Accordingly, the convex cone of the prices at time \(t,\) \(L^{+}\) is a subset of \(L_{p}^{+}( \Omega ,\mathcal{U},\mathcal{P}) =\{X\in L_{p}:X\geq 0\}.\) The authors are interested in the existence of a risk-neutral probability measure \(P^{0}\) equivalent to \(P\) such that for all \(X\in L_{p},\) the conditional expectations \(E^{0}[ X\mid B] \) are well defined elements of the \(L_{p}\)-space so that \(E^{0}[ R^{-1}X\mid B] \) is well defined for all \(X\in L_{p},\) thus giving the extension of the linear operator \(x( X)\), \(X\in L^{+}\) on the whole \(L_{p}\)-space. In the paper the authors suggest several new versions of extension theorems and their application to questions of the existence of a risk neutral probability measure. They are concerned with linear extensions which are monotone i.e., \(x( X^{\prime }) \geq x( X) \) for \(X'\geq X\) and also \(B\)-homogeneous, i.e., \( x( \lambda X) =\lambda x( X) \) for \(B\)-measurable multipliers \(\lambda \) such that \(\lambda X\in L_{p},\) properties which are motivated by the fact that they hold for the price operator. For the aforementioned class of operators they establish a kind of Hölder equality, both majorant preserving and sandwich preserving extension theorems and suggest some criteria for the existence of a risk neutral probability measure, which is the fundamental theorem of asset pricing.