A structural characterization of numéraires of convex sets of nonnegative random variables (Q1928543)

For a probability space \((\Omega,\mathcal F,\mathbb P)\) denote by \(L^0\) the equivalence classes of finite real-valued random variables on \((\Omega,\mathcal F),\) and let \(L_+^0\) be the set of all \(f\in L^0\) such that \(\mathbb P[f<0]=0\). With the metric \(d(f,g)=\mathbb E[\min\{|f-g|,1\}] \) (here \(\mathbb E\) stands for the expectation under \(\mathbb P\)), \(L^0\) becomes a complete metric space and \(L_+^0\) is closed in \(L^0\). An element \(g\) of a subset \(\mathcal C\) of \(L^0_+\) is called maximal in \(\mathcal C\) if \(\mathbb P[g\leq f]=1\) implies \(\mathbb P[g = f]=1\) for all \(f\in\mathcal C\). If \(\mathcal C\) is convex, then \(g\in\mathcal C\) is called strictly positive if \(\mathbb P[f>0, g=0]=0\) holds for all \(f\in \mathcal C\). A strictly positive element in \(\mathcal C\) is called a numéraire if \(\mathbb E[f/g : g>0]\leq 1\) for all \(f\in \mathcal C\). In Proposition 1.2 it is shown that a strictly positive element \(g\in\mathcal C\) is a numéraire iff there exists a \(\sigma\)-finite measure \(\mu\) on \((\Omega,\mathcal F)\), equivalent to \(\mathbb P\), such that \(\int g \, d\mu=\sup\{\int f \, d\mu : f\in\mathcal C\}\). Consequently, a numéraire \(g\) can be interpreted as a strictly positive support point of \(\mathcal C\). When \(\mathcal C \subset L^1(\Omega,\mathcal F,\mathbb P)\), then \(g\) is a support point of \(\mathcal C\) in the usual sense. In the main result of the paper (Theorem 1.4), the author gives a characterization of strictly positive elements which are numéarires, based on some ideas from financial mathematics.