Tankan Large All Industry Capex

Tankan Large All Industry Capex is the large-firm investment line of Tankan (Short-Term Economic Survey of Enterprises in Japan, a statutory business survey). The Bank of Japan (BOJ, Japan central bank) publishes it for large enterprises (capital of 1 billion yen or more) across 31 industries.

All industries means 17 manufacturing plus 14 nonmanufacturing industries, excluding financial, consulting, and holding firms, surveyed quarterly with about 9,000 to 10,000 firms. The survey covers firms capitalized at 20 million yen or more, and the headline has no threshold. The headline is the YoY (change versus prior fiscal year) % change in fixed investment (tangible assets including land), excluding software and R&D (research spending).

Forms go to sample firms about 1 month before release, with returns due about 2 weeks later and collection until 1 business day before release. March, June, September, and December surveys appear in early April, July, October, and mid-December at 8:50 a.m. JST (Tokyo time) over 2 days.

Day 1 brings summary figures and day 2 brings full details plus time series, alongside sales, profits, judgments, exchange rates, and inflation outlook. Each fiscal year is surveyed 6 times through next June, March data are less comparable, and corrections appear swiftly while delays can postpone release.

The total is a yen sum with no fixed weights, so large manufacturing, large nonmanufacturing, and the land, software, and research splits drive it.

• Large manufacturing fixed investment: newly listed tangible assets of large makers across 17 industries, about half the yen total.
• Large nonmanufacturing fixed investment: same tangible-asset concept for large nonmakers across 14 industries, the other half.
• Land purchasing expenses: new land purchases including developing and preparing, inside fixed investment and often volatile.
• Software investment: newly listed intangible software assets, reported alongside but excluded from the headline.
• R&D investment: income-statement research spending, reported alongside but excluded from the headline.
• Memo aggregates: supplemental totals adding financial firms and holding companies, outside core Tankan.

The December 13, 2024 survey showed headline fixed investment up 11.3% YoY on a 0.7% revision, yet its halves diverged sharply. Large manufacturing rose 16.0% but was cut 2.4% from September, while large nonmanufacturing rose 8.7% after a 2.5% upward revision.

Software rose 14.0% and research spending 6.7%, while land fell 6.1% with large manufacturing land down 30.6% inside the strong headline. December plans often fall by the next June actual survey as construction slips, so late strength needs caution. Large manufacturing sentiment stood at 14 and large nonmanufacturing at 33.

1. Stratify: Split firms into strata (subgroups) by industry, capital, and sales for 93 industry by size groups.
2. Expand: For each stratum, multiply the respondent average yen by population firms: e(i) = a(i)/n(i) x N(i). Here e is the stratum estimate, a is respondent yen, n is respondents, and N is population firms.
3. Aggregate: Sum stratum estimates into large manufacturing, large nonmanufacturing, and large all industries.
4. Year-on-year change: Divide the current fiscal-year yen total minus the prior-year total by the prior-year total, times 100. For example, stratum A with 4 respondents totaling 800 and 40 population firms gives 8,000. Stratum B with 6 totaling 900 and 60 firms gives 9,000. Totals of 17,000 versus 15,000 give (17,000 minus 15,000) / 15,000 x 100 = +13.3%.
5. Revision rate: Divide the current yen estimate minus the prior-survey estimate by the prior-survey estimate, times 100. It tracks yen shifts between surveys, not shifts in the YoY rate. For example, 17,000 versus a prior 16,500 gives (17,000 minus 16,500) / 16,500 x 100 = +3.0%.
6. Gaps, outliers, and seasonality: Fill missing firms with their most recent figures, neutralize extreme jumps, and apply no seasonal adjustment to capex.