# Data dictionary

**Dataset** Australian council-level household running cost dataset 2026-27
**Version** 2026-27+47f214b1
**Rows** 147, one per council
**Licence** CC BY 4.0, https://creativecommons.org/licenses/by/4.0/
**Canonical page** https://www.suburbcost.com.au/dataset/
**File** https://www.suburbcost.com.au/dataset/suburbcost-council-running-costs.csv

Attribution, copy this line:

> SuburbCost (2026). Australian council-level household running cost dataset, 2026-27. Version 2026-27+47f214b1. https://www.suburbcost.com.au/dataset/ Licensed CC BY 4.0.

## What one row is

One local government area, aggregated from the suburbs SuburbCost holds inside
it: 6,158 suburbs across 147 councils. A further
64 suburbs, all in New South Wales, carry no council assignment and
appear nowhere in this file.

**Suburb-level rows are not published.** This is a derived aggregate.

## Columns

| Column | Type | Meaning | Units |
| --- | --- | --- | --- |
| `council_name` | text | Council name as SuburbCost publishes it. Derived from councilSource. | n/a |
| `state` | text | QLD or NSW. Suburb coverage is these two states only. | n/a |
| `suburbcost_slug` | text | Resolves to https://www.suburbcost.com.au/council-rates/{slug}. | n/a |
| `suburb_count` | integer | Suburbs SuburbCost assigns to this council. Every median below is computed over exactly these suburbs. | count |
| `council_rate_aud` | integer | The modal stored rate for the council. Read council_rate_basis before using it. | AUD per year |
| `council_rate_basis` | enum | Which statistic council_rate_aud is. Four values, see Bases below. Averaging across them produces a meaningless number. | n/a |
| `council_rate_year` | text | Financial year the rate applies to. Derived, not read from the source string, which names two years for NSW. | n/a |
| `council_rate_source` | text | The source string verbatim, including the approved SRV percentage where one applies. | n/a |
| `council_rate_variant_count` | integer | Distinct rates stored inside this council. 1 for 144 councils. Logan sets 2 and Western Downs sets 3 by locality, so their single published figure is the modal one. | count |
| `median_total_house_aud_incl_est` | integer | Median of council + water + modelled commute + building insurance across the council's suburbs. Contains an estimated component, see Estimated fields. | AUD per year |
| `mean_total_house_aud_incl_est` | integer | Mean of the same. Published alongside the median because the two diverge sharply where a council mixes tapered and flat commute suburbs. | AUD per year |
| `median_total_unit_aud_incl_est` | integer | Median of councilUnit + water + strata + modelled commute + contents insurance. A different set of components from the house total, not a like-for-like comparison. | AUD per year |
| `median_water_aud` | integer | Median water and sewerage. Assumes roughly 190 kL a year. | AUD per year |
| `median_modelled_commute_drive_aud` | integer | Median modelled driving commute. Read commute_basis before using it, see Commute. | AUD per year |
| `median_insurance_house_aud_est` | integer | Median building insurance. Estimated for every row. | AUD per year |
| `median_strata_aud_est` | integer | Median body corporate, units only, 0 for houses. Estimated for every row. | AUD per year |
| `suburbs_rate_standin` | integer | Suburbs in this row whose own council rate is unknown and which carry this council's rate instead. See Substituted rates. | count |
| `suburbs_water_estimated` | integer | Suburbs whose water figure is a neighbouring utility's rather than their own. | count |
| `suburbs_flat_commute_allowance` | integer | Suburbs carrying the flat $18,200 regional driving allowance rather than a tapered commute. | count |
| `commute_basis` | enum | tapered, mixed or flat_allowance_only. See Commute. | n/a |
| `completeness` | enum | complete when all four flag columns are zero, the council stores one rate, and it has at least 3 suburbs. flagged otherwise. Nothing is excluded from this dataset, so filter on the flag columns rather than assuming a clean file. | n/a |

## Rounding

Every dollar figure is a whole number of Australian dollars, rounded **half
up**. This matters when reproducing the medians. 68 of the 147 councils hold an
even number of suburbs, so their median is the mean of two integers and can land
on a half dollar. 57 of the 876 median cells in this file do.

Python's `round()` and R's `round()` both use half-to-**even**, so
recomputing these medians in either language returns a figure one dollar lower
on those cells unless you round half up explicitly. That is a convention
difference, not a data disagreement.

## Bases, read this before computing anything across states

`council_rate_aud` is not one statistic. It is four, and they are not
interchangeable. A mean across the whole column is meaningless.

| `council_rate_basis` | Councils | What the figure is |
| --- | --- | --- |
| `qld_minimum_general_rate_ppr` | 20 | The **minimum** residential general rate for an owner-occupied principal place of residence. A floor. Most owner-occupiers pay more, scaled to land value. |
| `qld_median_property_rate` | 1 | Redland City only, which publishes a median-valued-property figure rather than a minimum. |
| `nsw_olg_average_residential_uplifted_by_ipart_peg` | 119 | The Office of Local Government average residential rate, uplifted by that council's IPART rate peg. An **average**, not a floor. |
| `nsw_olg_average_residential_uplifted_by_approved_srv` | 7 | The same, uplifted by an IPART-approved special rate variation above the peg. Separated so rate-capping research can isolate them. |

A Queensland minimum sits **below** a typical Queensland bill. A New South
Wales average sits **at** a typical New South Wales bill. Comparing the two
directly understates the real gap between the states rather than overstating
it. Waste, levies and separate charges are excluded throughout.

## Commute

`median_modelled_commute_drive_aud` is **modelled, not measured**, and the
model has two regimes.

Under 100 km from the relevant centre, the annual cost is
`distance x 2 x trips x 48 x $0.91`, where `trips = max(0.5, 4 - distance/20)`,
with a minimum costing distance of 1.5 km. At or beyond 100 km the figure is a
flat **$18,200**, which is 20,000 km a year of **all household driving**. That
is a different quantity from a commute and is deliberately not expected to line
up with the taper.

| `commute_basis` | Councils | Meaning |
| --- | --- | --- |
| `tapered` | 54 | No suburb uses the flat allowance. |
| `mixed` | 41 | Some do. The median may sit either side of the boundary. |
| `flat_allowance_only` | 52 | Every suburb uses it. The figure is **not a commute** for these councils. |

**Distance is straight-line, so the tapered figures understate real driving.**
Measured against 6,212 pinned road routes, median road distance is
**1.294x** straight-line. That factor is valid only for the 4,452 suburbs
measured to a state capital. For suburbs measured to a regional centre such as
Newcastle, Wollongong, Cairns or Townsville **no road routes exist and the
understatement is unmeasured**. No correction has been applied to this file.

The understatement runs in opposite directions. Metropolitan suburbs are
understated. Suburbs on the flat allowance are not understated at all, they are
measuring something else.

## Estimated fields

Columns ending `_est` are AI-estimated and not sourced from a provider.
Building insurance and body corporate are estimated for **every** row, so the
marker is in the column name rather than a per-row flag that would be constant.

The three `_incl_est` totals contain an estimated component. Every component
is published separately in this file, so an estimate-free subtotal can be
reconstructed by summing the columns you accept.

One caution on insurance. It is dominated by **cyclone exposure**, not flood or
fire. Its correlation with the published flood and fire ratings is r = 0.053.
Means run from $2,522 in New South Wales to about $7,500 north of 16 degrees
south. Do not read a high insurance figure as a flood finding.

## Substituted rates

Seven Queensland rows cover more councils than they name. 93 suburbs belong to
**21 Queensland councils that have no row in this file** and carry a
neighbouring council's rate instead. Douglas Shire, 34 suburbs, sits
inside Cairns Regional Council. Etheridge, Croydon, Cook, Mornington, Aurukun,
Kowanyama, Pormpuraaw and Yarrabah are also inside Cairns. South Burnett sits
inside Ipswich; Charters Towers, Flinders, McKinlay, Richmond, Hinchinbrook and
Carpentaria inside Townsville; Goondiwindi inside Toowoomba; Isaac and
Livingstone inside Mackay; Gympie inside Noosa and Sunshine Coast.

`suburbs_rate_standin` gives the count per row. Cairns Regional Council is the
extreme case at 62 of 120 suburbs. **Those councils are absent from this
dataset, not merely incomplete within it.**

## Water

181 suburbs across five Queensland councils carry a neighbouring utility's
water figure rather than their own, and in all five cases it is **every** suburb
in the council: Southern Downs, Mareeba Shire, Tablelands, Cassowary Coast and
Western Downs. `suburbs_water_estimated` equals `suburb_count` for those rows.

## Known limits, stated rather than buried

- **Nothing is excluded.** Only 50 of 147 rows carry no flag at all. A file
  restricted to clean rows would be a third of this one and would hide which
  councils were dropped, so every row is published and every gap is a column.
- **Medians are computed over all of a council's suburbs**, including flagged
  ones. The flag columns give the denominator so you can judge or refilter.
- Mosman Council holds one suburb, so its median is a single observation.
- 39 councils span more than one commute centre, so their median commute mixes
  trips to different destinations.
- Property prices, land valuations and capital growth are **not** in this
  dataset and are not published by SuburbCost anywhere.

## Provenance

Generated by `generate-dataset.js` from `data/suburbs.json`. The version
string is the data vintage plus a short hash of the rows, so it changes when a
figure changes and not merely because the generator ran.

Rate figures come from each council's own budget or revenue statement (QLD) and
from the NSW Office of Local Government "Your Council" release uplifted by the
relevant IPART determination (NSW). Water is from the named retailer where one
exists and is estimated for other regional NSW. Insurance and body corporate are
estimated. The commute model is described above.