... objects¹

The optimal decision $x=1/m$ is not unique, any decision satisfying the inequality $c_i x_i \ge a,\ i=1,...,m$ minimizes the expected utility function $U(y)$.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... applied²

Theoretically one can transform the case into the convex one by introducing randomization, however that is not acceptable in the practical insurance.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... objects³

It is assumed that each customer owns a single object of market value $z=1$ and that there are just two feasible insurance policies $x=1$ or $x=0$.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.