Adaptive Economic Operating System

Feedback, behaviour and disturbances

At the highest level, the system is represented as

where x_t is the socio-economic state, u_t is the vector of policy interventions (fiscal, monetary, macroprudential, regulatory), w_t is external disturbance the government does not control (energy and commodity shocks, geopolitics, pandemics, technology, climate, demographics), and θ_t captures behavioural and institutional structure. Almost all of the genuine difficulty is inside f(·); making that explicit is a large part of the point.

Two consequences follow directly:

1. People are not fixed transfer functions. Raising an income-tax rate does not mechanically raise τ × Y, because taxable income itself responds to the tax system: R(τ) = τ · Y_taxable(τ). Households and firms adjust labour supply, incorporation, investment location, saving and consumption in response to policy — differently across the population, not uniformly.

2. Because w_t is not chosen by government, policy needs to be robust, not merely optimal under expected conditions. A policy tuned to perform well in a specific forecast can perform badly once an energy shock, financial crisis or pandemic moves the system somewhere the forecast did not anticipate. Designing for a range of plausible disturbances, rather than a point forecast, is treated as a first-class requirement rather than a refinement to add later.

The remainder of this specification is largely about designing sensible forms for u_t (the policy controllers) against a f(·) that is explicitly heterogeneous, credit-endogenous and behaviourally responsive — and about measuring x_t well enough to know whether u_t is working.