Capital Expenditure MoM

Private New Capital Expenditure and Expected Expenditure, Australia measures actual and expected new private-business spending on buildings and equipment by industry, called capex. The Australian Bureau of Statistics (ABS) runs a quarterly Survey of New Capital Expenditure on a random sample of about 10,000 units. Units are stratified by industry, state and derived employment size, plus large units outside the frame.

It covers private businesses in selected industries such as mining, manufacturing, construction, retail and services. It excludes agriculture, the public sector, household dwellings, land and used buildings. It is not economy-wide investment or the investment part of the National Accounts. It gives no monthly reading, so quarter-on-quarter change is the shortest interval.

Returns are completed in 8 or 9 weeks after quarter end on a management-accounts basis through an online survey. March-quarter returns close during April and May. The release is Thursday about 8 to 9 weeks after quarter end at 11:30am Sydney time. Each release provides key statistics, charts, data tables, Excel downloads and methodology, with state expectations only in the December-quarter release.

Initial estimates are preliminary and revised in the next release, while adjusted, trend and volume series are revised as data arrive. Trend estimates for the last 3 quarters are routinely revised because asymmetric averages apply at the end. A new base year arrives with the September-quarter release and the reference year advances.

Two asset types drive the total, with industry and state splits and an expectations ladder.

• Buildings and structures: industrial and commercial buildings, roads, bridges, pipelines and mine development, excluding land, used buildings and speculative projects.
• Equipment, plant and machinery: plant, machinery, vehicles, electrical apparatus, office equipment and furniture, including goods imported for the first time.
• Industry and geography splits: actual spending is spread across industries and states by each unit's main activity, and state expectations follow spending shares.
• Expectations ladder: 7 estimates per financial year (12 months ending 30 June), from actual spending plus expected remainder to the final.

March quarter 2024, released 30 May 2024, showed how parts can offset. Total new capital expenditure was $40,487m, up 1.0% on the quarter and 5.5% on March quarter 2023. Buildings fell 0.9% to $21,449m while equipment rose 3.3% to $19,038m, and trend rose 0.6% to $40,402m.

Estimate 2 for 2024-25 was $155.4b, 6.8% above Estimate 1. A small headline move can hide offsetting moves; early expectations are only a starting point.

1. Collection: Each sampled unit reports three figures quarterly: actual spending (Act) plus short-term (E1) and longer-term (E2) expectations from its own business accounts.
2. Editing and completion: Data are checked at unit and total level. Missing replies are estimated and later replaced, with additions for large spenders outside the sample and new businesses.
3. Aggregation: Totals use number-raised estimation, which grosses up each group by the inverse of its selection chance. For example, $800m reported with a factor of 4 contributes $3,200m.
4. Seasonal adjustment: Original totals become seasonally adjusted totals removing normal seasonal patterns. Total growth uses SA, the adjusted value, as g = 100 * (SA_t - SA_{t-1}) / SA_{t-1}. For example, $40,000m with a factor of 0.99 gives about $40,404m, and after $40,061m that is 0.9%.
5. Trend: Trend is a 7-term Henderson moving average of the adjusted series. End values use asymmetric weights, so the last 3 quarters are routinely revised.
6. Volumes: Volumes are annually reweighted chain Laspeyres indexes, linked measures valued at prior-year prices. Yearly growth is linked to a reference-year level that advances every September, and volume parts do not add up.
7. Expectations adjustment: Expected spending is scaled by the realisation ratio R, past actuals over past expectations. For example, $150.0b expected with R of 1.05 gives $157.5b, though early estimates predict weakly.