Optimal Policy.

Policy under commitment: jointly optimises ten cash rates, subject to rate and quarterly-change limits.

Systematic policy: fits a rule’s inflation, unemployment, momentum and inertia coefficients using the Gross & Leigh (2022) rule. The same rate limits apply. Inertia ranges from 0–1; other coefficients from 0–5.

Multiple starting points reduce search risk but do not guarantee a global optimum. The restricted rule can perform worse than the baseline. Inertia and the smoothing weight are separate.

Loss sums squared inflation and unemployment gaps and quarterly rate changes, without discounting. Default weights: 1, 1, 0.5. Inflation target: 2.5%, applying the headline CPI target to trimmed mean.

The unemployment target defaults to 4.89%. Changing this target leaves the underlying NAIRU and model responses unchanged. The cash-rate anchor is June 2026’s 4.3%. March and September forecasts are interpolated.

The rule’s neutral real rate defaults to 1%. Initial unemployment momentum uses the June anchor for the missing prior quarter.

No uncertainty or loss beyond December 2028 is included.

August 2026 SMP forecasts with public MARTIN equations. Each quarter’s cash-rate equation is replaced; other equations remain. Responses use the saved 2019 model baseline, without re-estimation. This linear approximation is independent of the RBA and less reliable for large changes.

DINGO: uses the baseline solution released with Fink & Hambur (2026). Monetary-shock responses are inverted to deliver each chosen cash rate, allowing for earlier shocks. Four quarterly inflation responses form year-ended trimmed mean inflation.

DINGO unemployment is a proxy: Δ(u − u*) = −0.8 × Δ output gap. The model’s GDP-level response approximates the gap change, holding potential output and NAIRU fixed. The coefficient follows RBA evidence: a 1.25% output gap corresponds to roughly a −1 pp unemployment gap. This is a calibration, not an estimated DINGO unemployment equation.

DINGO treats quarterly interventions as successive surprises under the original expectations rule. Both policy methods are conditional IRF exercises, not a re-solved rational-expectations commitment or new-rule equilibrium. Optimising the whole rate path does not make future changes anticipated.

Animation reveals the committed path sequentially; systematic policy moves together. Intermediate frames are visual transitions. Displayed loss is recalculated and can temporarily rise.

Market rates come from my cash-rate repository on loading and hourly thereafter. Saved copies are labelled. ASX futures imply monthly averages, shown at month-end; risk premia and forecast dates differ. No extrapolation. Market rates do not enter the optimisers.