Skip to content

DCT Lexicon · Governing Dynamics

Dynamic Risk Measures

A dynamic risk measure assesses tomorrow's uncertain position with what is known today, and stays consistent as information arrives: what is acceptable tomorrow in every state is acceptable today.

DCT Governing Dynamics

In depth

Formal statement

A conditional risk mapping ρt:Lp(Ft+1)→Lp(Ft)\rho_t : L^p(\mathcal{F}_{t+1}) \to L^p(\mathcal{F}_t) is coherent if monotone, translation-equivariant, convex and positively homogeneous. Time consistency yields the nested form.

ρ0,T(Z)  =  ρ0(ρ1(⋯ρT−1(Z)))\rho_{0,T}(Z) \;=\; \rho_0\Big(\rho_1\big(\cdots \rho_{T-1}(Z)\big)\Big)AV@Rα(Z)  =  inf⁡c∈R{c+1α E[(Z−c)+]}\mathrm{AV@R}_\alpha(Z) \;=\; \inf_{c \in \mathbb{R}} \Big\{ c + \tfrac{1}{\alpha}\, \mathbb{E}\big[(Z - c)_+\big] \Big\}

DCT Governing Dynamics: the element's drawer (/governing-dynamics)

Dual representation and time consistency

A coherent risk measure is a worst-case expectation over a convex set of probability measures; it is time-consistent exactly when that set is rectangular across stages, which makes risk-averse dynamic programming possible.

ρ(Z)=sup⁡Q∈Q  EQ[Z]\rho(Z) = \sup_{Q \in \mathcal{Q}}\; \mathbb{E}_Q[Z]

DCT Governing Dynamics: the element's drawer (/governing-dynamics)

In DCT

A board appetite stated as "at most 5% of paths breach" is a static chance constraint; a nested risk measure is how that appetite is enforced at every gate as the programme unfolds, so that tomorrow's re-certification never contradicts today's.

DCT Governing Dynamics: the element's drawer (/governing-dynamics)

Limits

Value-at-Risk is neither convex nor time-consistent. Nested AV@R is time-consistent but can be far more conservative than a single AV@R on the total, increasingly so over long horizons.

DCT Governing Dynamics: the element's drawer (/governing-dynamics)

Go next

  • Chance Constraints and Certificates: Chance constraints are the probabilistic face of risk limits (Pathway 9).
  • Wasserstein Distributionally Robust Optimization: The dual set of a coherent risk measure is an ambiguity set.
  • HJB Equations and Viscosity Solutions: Risk-sensitive control modifies the Hamiltonian.

DCT Governing Dynamics: the element's drawer (/governing-dynamics)

References

  • Artzner, Delbaen, Eber & Heath (1999)
  • Rockafellar & Uryasev (2000)
  • Ruszczyński (2010), Risk-averse dynamic programming for Markov decision processes

DCT Governing Dynamics: the element's drawer (/governing-dynamics)

In the book: Vol. II, Ch. 9: Stochastic Enterprise Optimization · Vol. I, Ch. 12: Enterprise Risk, Resilience, and Robustness Architecture

Related: Chance Constraints and Certificates · Wasserstein Distributionally Robust Optimization · HJB Equations and Viscosity Solutions

GD.16