A closed oriented Riemannian manifold with negative sectional curvatures has the property that all its geodesics have Morse index zero.
Is there a known counterexample to the "converse": if (M,g) is a closed oriented Riemannian manifold (Edit: assumed to be nondegenerate) all of whose geodesics have Morse index zero then M admits a (possibly different) metric g' with negative sectional curvatures?
Edit: Motivation for asking this (admittedly naive) question is that Viterbo/Eliashberg have proved that a manifold with a negatively curved metric cannot be embedded as a Lagrangian submanifold of a uniruled symplectic manifold. Actually their proof only seems to use the existence of a nondegenerate metric all of whose geodesics have Morse index zero. I wondered if that was known to be strictly weaker.