Australian Housing Affordability Dashboard

Interactive R/Shiny dashboard for Australian housing affordability and cost-pressure analysis. The app combines saved ABS, SIH and RBA-derived CSV outputs with Plotly charts for prices, rents, official housing burden measures, labour-market context, housing supply and stylised market-entry scenarios.

Live web app here: https://anthonypuggs-household-affordability-dashboard.share.connect.posit.cloud/

Run The App

Restore the pinned R package environment from renv.lock:

Rscript -e "renv::restore()"

renv.lock pins package versions for reproducible dashboard, pipeline and test runs. In plain terms, renv.lock pins package versions so another machine can restore the same R package set. The project activation file .Rprofile loads renv automatically when R starts from the repository root.

If renv is unavailable, install it first:

install.packages("renv")

As a manual fallback, install the direct runtime and pipeline packages:

install.packages(c(
  "shiny", "bslib", "ggplot2", "plotly", "dplyr", "tidyr", "purrr",
  "stringr", "scales", "readr", "readxl", "readabs", "lubridate",
  "httr", "rlang", "watcher"
))

From the repository root, start the dashboard with:

Rscript -e "shiny::runApp('.')"

The app reads pre-generated CSV files from data/, so it can run without refreshing live ABS or RBA inputs.

The dashboard uses bslib-native dark/light mode through Bootstrap 5. It uses local system font stacks rather than Google-hosted fonts, avoiding third-party font fetches during launch or deployment.

The Shiny interface uses a public-policy report UI system in R/ui_style_system.R. Page headers, calmer KPI tiles, chart cards and source notes are helper-backed so modules share the same hierarchy, spacing and local/system-font treatment.

Refresh The Data

Run the full data pipeline from the repository root:

Rscript pipeline/05_driver.R

The pipeline parses local ABS Survey of Income and Housing workbooks under resources/ABS_data/, retrieves public ABS/RBA time series, derives affordability and cost-pressure indicators, and validates the generated outputs.

Pipeline runs use per-stage output gates from R/pipeline_contracts.R before continuing to the next stage. The same helper records the fixed external-source manifest for ABS catalogue/table calls, ABS SDMX CPI endpoints and RBA F-series tables.

Data Model

Main dashboard CSVs live in data/:

• abs_timeseries.csv: long-format ABS macro, CPI, price, labour and supply series using date | value | series | series_id | category | unit | frequency.
• rba_rates.csv: long-format RBA cash and mortgage-rate inputs using the same time-series schema.
• rba_*_raw.csv: normalised RBA source-cache artefacts kept rectangular for reproducible parsing; rba_rates.csv is the dashboard-ready RBA output.
• affordability_indices.csv: derived cost-pressure indicators using date | value | indicator | geography | unit | frequency.
• sih_*.csv: parsed ABS Survey of Income and Housing tables for official housing cost, burden and NHHA rental-stress measures.
• sih_estimate_quality.csv: SIH sampling-error metadata for selected workbook tables, including 95% margin of error values and relative standard error flags. Users should interpret with caution when estimates have RSE from 25% to 50%; estimates above 50% are too unreliable for general use.
• SIH estimate outputs are guarded by workbook benchmark checks in R/sih_benchmarks.R, covering key rows from ABS SIH Files 4, 5, 8 and 13 so sampling-error sections do not contaminate the main estimate CSVs.

Official SIH/NHHA measures should be interpreted separately from modelled market-entry indicators. Mortgage serviceability, deposit-gap and calculator outputs are stylised scenarios, not official ABS measures or lender assessments.

R/market_entry_scenarios.R centralises app-only market-entry scenario calculations for mortgage repayments, assessed-rate sensitivity, deposit saving time and expense-adjusted serviceability ratios. Assessment buffer, deposit, implied LVR, loan-term and expense inputs are sensitivity assumptions, not a lender assessment. The serviceability chart uses AWE individual earnings as the income proxy; savings-rate assumptions remain calculator-only because they affect deposit saving time rather than repayment serviceability.

Methodology Metadata

R/indicator_registry.R is the source of truth for derived indicator formulas, source series, units, interpretation direction and caveats. The registry documents the current formulas used by the pipeline and dashboard; it does not make stylised market-entry measures official ABS measures or lender assessments.

R/pipeline_contracts.R documents the pipeline stage contracts and fixed external-source manifest. It keeps the release surface explicit: each data-producing stage has required CSV outputs, and ABS/RBA source surfaces are listed separately from the saved dashboard data.

The Shiny app includes a Methodology page backed by this registry. It shows the formula, source series, interpretation direction and official/stylised status for each derived affordability indicator.

R/methodology_module.R, R/affordability_module.R, R/rental_market_module.R, R/housing_supply_module.R, R/price_trends_module.R, R/geographic_affordability_module.R, R/market_context_module.R and R/overview_module.R are the Shiny page modules. They keep the registry-backed methodology/provenance page, Affordability page, Rental Market page, Housing Supply page, Price Trends page, Geographic Affordability page, Market Context page and Overview page isolated while broader component-level refactors remain incremental.

plot_setup.R is a thin compatibility entrypoint over focused support modules for CSV loading, dashboard formatting, theme helpers and app-ready precomputed series. It keeps the existing global object names available to page modules while reducing the setup file's direct responsibility.

R/chart_builders.R is the page-level chart-construction helper surface. It keeps ggplot builders testable outside Shiny while modules retain input handling, validation, caching, SIH quality joins, Plotly conversion and deliberately Plotly-specific post-processing such as right-side annotations.

For the practical chart-editing workflow, see CHART_BUILDER_WORKFLOW.md.

For release smoke testing of the live Shiny interface, see docs/ui_smoke_checklist.md. The project uses static base-R UI smoke contracts and the Codex in-app browser checklist, with no new browser-testing dependencies.

The Housing Supply page keeps Building Approvals readable by filtering the ABS approval series with state, building-type and sector controls; the default view compares total-sector total approvals for New South Wales and Victoria.

The Geographic Affordability page is an SIH-only, geography-aligned view. It compares state, lower-income state and greater-capital-city/rest-of-state SIH estimates where the housing-cost numerator and income or household denominator are measured within the same geography; it does not construct state or capital-city market-entry indexes from national wage, RPPI, AWE, WPI or CPI-rent proxies.

Survey charts that have matching SIH estimate-quality metadata display compact reliability markers (†) and include relative standard error or margin-of-error details in Plotly hover text. Where available, visible error bars and interval hover text use 95% margin-of-error metadata from data/sih_estimate_quality.csv.

KPI colours encode economic interpretation as better, worse or neutral/contextual rather than raw up/down movement. R/visual_semantics.R centralises these classes and chart palettes so affordability-worsening increases are not shown as favourable.

The Methodology page also provides an on-demand Markdown download generated by R/provenance_report.R. The download combines the registry-backed methodology table with saved data/*.csv inventory metadata, without writing a persistent report into the repository.

Verification

The project uses lightweight base-R tests. Useful checks from the repository root are:

Rscript tests/test_pipeline_outputs.R
Rscript tests/test_app_output_ids.R
Rscript tests/test_kpi_change_labels.R
Rscript tests/test_app_method_text.R
Rscript tests/test_methodology_page.R
Rscript tests/test_methodology_module.R
Rscript tests/test_affordability_module.R
Rscript tests/test_rental_market_module.R
Rscript tests/test_housing_supply_module.R
Rscript tests/test_housing_supply_legend_contracts.R
Rscript tests/test_price_trends_module.R
Rscript tests/test_geographic_affordability_module.R
Rscript tests/test_geographic_affordability_data_contracts.R
Rscript tests/test_market_context_module.R
Rscript tests/test_overview_module.R
Rscript tests/test_provenance_report.R
Rscript tests/test_sih_quality_helpers.R
Rscript tests/test_sih_uncertainty_intervals.R
Rscript tests/test_sih_estimate_quality.R
Rscript tests/test_sih_workbook_benchmarks.R
Rscript tests/test_market_entry_scenarios.R
Rscript tests/test_serviceability_scenario_controls.R
Rscript tests/test_visual_semantics.R
Rscript tests/test_chart_builders.R
Rscript tests/test_plot_setup_extraction.R
Rscript tests/test_plotly_helpers.R
Rscript tests/test_app_plotly_cache_contracts.R
Rscript tests/test_ui_style_system.R
Rscript tests/test_ui_smoke_contracts.R
Rscript tests/test_responsive_ui_contracts.R
Rscript tests/test_rental_market_mobile_contracts.R
Rscript tests/test_rental_market_interaction_contracts.R
Rscript tests/test_theme_infrastructure.R
Rscript tests/test_rba_raw_cache_hygiene.R
Rscript tests/test_public_release_hygiene.R
Rscript tests/test_project_root_paths.R

Before publishing, run the release-readiness checklist:

Rscript -e "source('R/release_checklist.R'); validate_release_checklist()"

Release checklist warnings can be acceptable for known data vintage, while failures block public release until fixed.

For a quick source check:

Rscript -e "source('plot_setup.R'); source('app.R'); cat('APP_SOURCE_OK\n')"

Notes

• pipeline/05_driver.R is the canonical data-refresh entrypoint.
• app_old.R, _check_cpi.R and save_plots.R are legacy or manual-support scripts and are not part of the production app launch path.
• HOUSING_DASHBOARD_ROOT can be set to the repository path when launching scripts from unusual working directories.