Australian Economic Analysis / MARTIN
Government spending,
inflation & interest rates.
Compare government consumption and investment, with a temporary or permanent increase in spending.
Four fiscal experiments
The same extra spending in each case, using public MARTIN (2019). The cash rate responds through the model’s monetary policy rule.
Negative values model spending cuts.
Transitory: higher spending for four quarters, then back to baseline. Permanent: a higher spending share maintained throughout the ten years shown. This is a level shift, not faster spending growth every year.
Loading model responses…
The spending paths overlap by construction. Changes relative to an unchanged-policy baseline, not forecasts of today’s economy. Inflation is year-ended; cash-rate changes are in percentage points. One percentage point equals 100 basis points.
Peak effects and quarterly values
Peak means the largest absolute deviation within the displayed horizon, with its sign retained. Quarter 1 is the first quarter of the shock.
| Scenario | Inflation (pp) | Cash rate (pp) | GDP (%) |
|---|
What the comparison shows
In this released calibration, higher consumption initially raises inflation and the cash rate. Maintaining the increase leaves a more persistent inflation response than withdrawing it after one year.
The investment result needs care: GDP, inflation and the cash rate subsequently fall below baseline. The source equations give investment a particularly strong import-demand effect. This is a result of the released calibration and baseline, not evidence that public investment generally lowers inflation, and not a modelled productivity dividend.
How the effects travel through MARTIN
Extra purchases lift demand and employment. A tighter labour market feeds into wages and prices. The policy rule raises the cash rate in response to inflation and labour-market conditions; borrowing costs and the exchange rate then restrain private demand. Import demand absorbs part of the spending increase. These responses interact, so the inflation path need not rise immediately or stay above baseline in every quarter. RBA model equations.
Compare like with like: 1% more government consumption and 1% more government investment are different-sized stimulus packages. Here both receive the same additional amount relative to baseline GDP. Consumption and investment have different import weights in the released equations.
A permanent spending increase does not mean permanently higher inflation. Monetary feedback and displacement of other spending shape the adjustment. Ten years of responses cannot establish the eventual steady state.
The NDIS question
Households use the services. Government pays.
NDIS-funded services are included in government consumption. For example, the ABS explicitly lists NDIS spending among the drivers of government consumption in its December 2025 national accounts. They are not missing from GDP simply because they are missing from household consumption expenditure.
| Measure | Where government-funded NDIS services fit |
|---|---|
| Government final consumption expenditure | Includes the government-financed services supplied to households. |
| Household final consumption expenditure | Excludes the government-funded portion; household out-of-pocket spending is separate. |
| Household actual final consumption | Adds social transfers in kind, including NDIS services, to household expenditure. |
| GDP | Records the services once. Moving between expenditure and actual-consumption presentations does not create additional output. |
The ABS identifies the NDIS as a social assistance benefit in kind in its adjusted disposable income accounts. That differs from an unrestricted cash benefit, which changes household income and affects demand when recipients spend it.
How to represent it in MARTIN
The natural aggregate mapping is an increase in real government consumption, gc, for additional NDIS-funded services. That is an application of the accounting classification to MARTIN, not a separately estimated NDIS equation. Do not also add the same services to household consumption: that would count the initial demand twice.
Use the net additional service volume relative to the baseline. A larger nominal budget can reflect higher prices, transfers between governments, or replacement of existing services rather than more real consumption. A 10% increase in NDIS spending is not a 10% increase in all government consumption.
MARTIN has no separate disability-services sector or NDIS programme. These generic shocks cannot isolate care-worker shortages, administered prices, changes in participants’ or carers’ labour supply, substitution for privately funded care, or the programme’s welfare benefits. Nor is an increase in service costs mechanically a one-for-one increase in household CPI. Programme-specific analysis needs those additional assumptions.
Model vintage, experiment design and sources
Author’s calculations using the public model accompanying Ballantyne et al. (2019), MARTIN Has Its Place: A Macroeconometric Model of the Australian Economy, RDP 2019-07. This is a local first-order translation of the 217 stored equations around the workfile’s saved 2019Q1–2028Q4 baseline, not the current internal RBA model or a forecast calibrated to current conditions. Some proprietary series were replaced with simulated data in the RBA public release.
Each quarter’s extra real spending equals the selected percentage of that quarter’s baseline real GDP. We replace the two government-spending equations with prescribed paths: one rises, the other stays at baseline. The temporary path returns exactly to baseline from quarter 5; the permanent path does not reverse within the simulation. It is not a one-off residual shock allowed to decay under the original fiscal equations.
All other equations remain active, including the cash-rate rule. Inflation expectations, the neutral rate and trend supply remain on their baseline paths. The model does not give households advance knowledge of the spending expiry: temporary and permanent paths coincide for their first four quarters. This is not a forward-looking announcement experiment.
The saved import-demand weight is about 0.282 for government investment and 0.086 for consumption. At the initial baseline, the direct long-run import derivatives (holding other demand and prices fixed) are about 1.29 and 0.10 real units per additional unit of spending respectively. The investment value above one helps explain its unusual negative responses. These are properties of the released equations, not independently measured import contents of a new project.
No explicit financing package, government-debt feedback or future tax adjustment is added. Results scale linearly with shock size; large changes and programme-specific effects require more detailed modelling. The finite nonlinear equations were checked against small-shock responses, including the temporary spending withdrawal. Native EViews execution was not performed. No confidence bands are estimated.