Baseline projections

Baseline projections are the reference path against which experiment projections are compared. They combine the model database, exogenous projections, ordered baseline simulation layers and model-wide tuning.

Related YAML documentation: model configuration, baseline tuning, design YAML and setup YAML.

Build 199 baseline files

<version>/199/
├── model_configuration.yaml
├── data/
│   ├── baseline_design.yaml
│   ├── baseline_exogenous_projections.csv
│   └── [other model data]
└── simulations/
    └── baseline/
        ├── setup.py
        ├── setup.yaml
        ├── run_baseline.py
        └── chartpack.yaml

The baseline design belongs in data/baseline_design.yaml. Do not put a second design in simulations/baseline/.

Baseline design

The design orders the simulation layers used to construct the baseline:

schema_version: 1
layers:
  - name: Growth and efficiency adjustments
    data_file: baseline_exogenous_projections.csv
    event_year: 2023
  - name: Monetary-policy adjustments
    data_file: baseline_INTX_projections.csv
    event_year: 2023

See design layers and layer ordering.

Standard growth and efficiency adjustments

Baseline projections normally include exogenous adjustments for labour-force growth, labour-augmenting productivity and autonomous energy efficiency. The source CSV files are described in the build 199 data-file guide. The economic basis for the projections is explained in the catch-up growth model and autonomous energy efficiency improvements pages.

Model-data preparation combines these inputs into two generated CSV files:

• labor_augmenting_technical_change.csv contains labour-augmenting technical change for all sectors and regions; and
• baseline_exogenous_projections.csv contains effective labour productivity growth, autonomous energy efficiency improvements and implied potential-output growth.

The baseline design can list several adjustment layers. G-Cubed applies them in their YAML order, which is useful when different types of baseline adjustment need to remain separate and reviewable.

Baseline tuning

Model configuration owns baseline-tuning choices:

• base-year real-interest-rate defaults and regional overrides; and
• monetary-policy defaults and regional overrides.

Generated baseline inputs are replaced during the relevant model-data update unless their model-configuration preservation flags are true.

Use projections.preserve_labor_augmenting_technical_change_file and projections.preserve_baseline_exogenous_projections_file only when the corresponding generated CSV has deliberately become a maintained input. With the normal false setting, change the source data and rerun model-data preparation rather than editing the generated file.

Potential-output projections

Potential-output growth is exogenous in the baseline and must remain consistent with the labour-force and productivity assumptions from which it is derived. An inconsistent path can distort experiment results because G-Cubed monetary-policy rules commonly respond to deviations from potential output.

Setup and run ownership

The optional baseline setup.yaml generates declared chartpacks and standard artefacts. It does not generate or configure run_baseline.py. The maintained run script selects resources, solves the model and writes results. It continues to work after the setup files are removed.

Run setup.py only when declared artefacts need to be regenerated. Run run_baseline.py to produce baseline projections. When model equations or model data change, use the maintained run script’s force-resolve setting rather than deleting the entire results directory.