Adaptive Economic Operating System

Stability and robustness

Every adaptive mechanism in this specification carries an explicit warning against changing too quickly or too unpredictably. Closed-Loop Fiscal Control requires bounded rate changes (|τ(t+1) − τ(t)| ≤ Δτ_max), smoothing, forward guidance and minimum policy review periods — the same discipline any control system needs to avoid oscillation, and that businesses and households need to plan around multi-year investment and consumption decisions.

Multiple controllers acting on the same coupled system can also counteract one another — fiscal expansion meeting monetary tightening is the clearest example, but similar interactions can occur between environmental taxation, macroprudential limits and public spending. Coordinating these controllers so they do not fight each other, or destabilise the system between them, is treated here as an open design problem rather than something assumed to resolve itself.

Finally, because government does not control external disturbances (w(t): energy shocks, geopolitics, pandemics, technology change), policy needs to produce acceptable outcomes across a range of plausible conditions, not only the one a specific forecast anticipated. The longer-run direction implied by this — periodically re-forecasting system trajectories under candidate interventions and re-planning as data improves, i.e. something like model predictive control — is identified here as a plausible future evolution of the "policy operating model" design decision, not something this specification claims is already fully worked out. The position taken today (see Design Decisions) is feedback control: measured error driving bounded, smoothed policy adjustment. Predictive/adaptive control is future research, not a component already specified.