We introduce the concept of a risk form, which is a real functional of two arguments: a measurable function on a Polish space and a measure on that space. We generalize the duality theory and the Kusuoka representation to this setting. For a risk form acting on a product of Polish spaces, we define marginal and conditional forms and we prove a disintegration formula, which represents a risk form as a composition of its marginal and conditional forms. We apply the proposed approach to two-stage stochastic programming problems with partial information and decision-dependent observation distribution.Proof. Using the support property twice, the translation property, and the normalization property, we obtain the chain of equations:This property was called state-consistency in [9]. Essential role in our analysis will be played by the increasing convex order (the counterpart of the second order stochastic dominance, when smaller outcomes are preferred).Here, [a] + = max(0, a).This concept allows us to consider risk forms consistent with the increasing order.Definition 2.4. A risk form ρ : (X ) × P(X ) → Ê is consistent with the increasing convex order, ifEvidently, consistency with the increasing convex order implies monotonicity and law invariance. We call two functions Z,V ∈ (X ) comonotonic, ifDefinition 2.5. A risk form ρ : (X ) × P(X ) → Ê is comonotonically convex, if for all comonotonic functions Z,V ∈ (X ), all P ∈ P(X ), and all λ ∈ [0, 1], ρ[λ Z + (1 − λ )V, P] ≤ λ ρ[Z, P] + (1 − λ )ρ[V, P].3