Recording Land in the National Balance Sheet

Summary

"Recording Land in the National Balance Sheet" (document LG/15/11/1) is an information paper by Andrew Cadogan-Cowper and Peter Comisari of the Australian Bureau of Statistics' Centre of Environment and Energy Statistics, presented to the 15th Meeting of the London Group on Environmental Accounting — the UN-linked expert group that develops the System of Environmental-Economic Accounting (SEEA) — in Wiesbaden, Germany, 30 November–4 December 2009. It reports the ABS's two-decade experience compiling land values for Australia's national balance sheet and the implications of the (then-new) 2008 System of National Accounts (SNA) for how the revised SEEA should treat land. Fetched and read in full (13 pages) from the UN Statistics Division's own hosted copy.

This is a technical methodology paper for national statisticians, not an economics paper and not a Georgist source — it takes no position on land value taxation. Its wiki relevance is as a rare primary account, from official statisticians, of why land's share of national wealth is difficult to measure and chronically under-published, and of the two competing valuation methods statistical agencies actually use.

The Core Argument / Findings

Where land value is actually measured, it is enormous: in Australia's national balance sheet, land represented 45–49% of total non-financial assets across 2002–2008 (chain volume terms) and 90% of all environmental-asset value in 2008. Yet a 2007 OECD survey (Aspden, 2008) found only twelve OECD countries produced any land-value stock estimate at all, often incomplete or intermittent even among these. Where comparable figures existed, land's share of non-financial assets ranged from 21% (Denmark, 2001) to 42% (France, 2008) and 30% (Canada, 2007) — the same order of magnitude as Australia's. The paper's diagnosis: "the principal problem appears to be in generating quality estimates for non-produced assets, particularly for land," because land is thinly traded and structures are typically sold together with it, making the land component hard to isolate.

The paper documents two valuation methods via Australia's own history. The direct/appraisal method — state Valuers-General (VGs) assessing land at market prices for revenue purposes — proved systematically unreliable: a 2005 New South Wales Ombudsman investigation found VG valuations relied on outdated, unadjusted sales data, used "mass valuation" techniques not regularly updated, and operated under an incentive "to value conservatively to avoid disputes and potential litigation with land owners." When the ABS compared VG-based estimates against an independent Reserve Bank of Australia (RBA) estimate, it found the VG figures substantially understated land value and switched methods. The RBA's approach is the residual method: multiply the mean market value of dwellings (house and land combined) by a census-based dwelling count to get total housing-stock value, then subtract the separately estimated net capital stock of structures — built via a perpetual inventory model (PIM) tracking construction spending and asset lives — leaving land value as the remainder. The resulting 2006–07 estimate: roughly A$1,265 billion of land value against a combined land-and-dwellings total of A$3,604 billion.

A further problem is measuring land's volume (real quantity) separately from its current-price value, since national accounts must distinguish inflation from real asset change. Because land's physical area barely changes, the paper treats location itself as a "quality," and therefore volume, factor: the 2008 SNA holds that "whenever a difference in price is found between two goods and services that appear to be physically identical there must be some other factor, such as location, timing or conditions of sale, that is introducing a difference in quality" (para 15.67). The ABS's actual method for splitting price change into real and inflation components is admittedly crude: rural land is assumed zero volume growth by convention, while urban land's volume growth is proxied as half the growth rate of overlying non-dwelling construction (one-third for land under dwellings) — an approximation, not a direct measurement.

Finally, the 2008 SNA now records land itself (a non-produced natural-resource asset, AN211) separately from "land improvements" (a produced, depreciable fixed asset, AN1123) and from ownership-transfer costs (AN116, also separately depreciable) — a change the paper recommends the SEEA adopt. Under the older 1993 treatment, transfer costs were folded into the land asset, so recorded land value could rise simply from repeated selling — "a result that is both counter-intuitive and also inconsistent with the SEEA objective to appropriately measure natural capital."

Relation to the Georgist Case

The paper corroborates, from an official statistical agency rather than an advocacy source, the wiki's recurring finding that land is a very large share of national wealth wherever measured — the same range (roughly 20–50% of non-financial assets) documented for the US by Larson (2015) and globally by McKinsey Global Institute. Its distinctive contribution is methodological: a primary account of why that share is chronically under-published and contested. Only a minority of OECD countries even attempt land-value estimates, and the more intuitive method — direct appraisal — is demonstrably prone to systematic undervaluation, while the residual method is only as reliable as the separate structure-value estimate feeding it. This is useful groundwork for the wiki's land-data lane: it explains, from the assessor's side, the measurement difficulty an LVT's administration must overcome, and gives historical context for the residual-versus-hedonic debate that later, more granular efforts — the FHFA's tract-level project and Albouy, Ehrlich, and Shin's transaction-based index — moved beyond. The paper itself is silent on LVT and takes no normative position; the Georgist relevance is the wiki's own extrapolation from a technical measurement account, not a claim in the source.

Nuances and Limits

• A conference information paper, not a peer-reviewed study. A methodological report by two national-accounts statisticians for a UN-linked technical working group, presenting institutional experience rather than hypothesis-tested research.
• Single-country focus. The empirical core is Australia; comparison figures for Canada, France, and Denmark are drawn from a cited but not independently reproduced OECD survey, so this paper alone cannot verify those countries' methodologies.
• The headline 45–49% figure is itself a product of the residual method it describes, inheriting whatever error exists in the ABS/RBA's separately estimated dwelling stock — a dependency the paper flags in explaining why earlier estimates were revised.
• The volume measurement techniques are explicitly approximate — proxying urban land's real growth via a fraction of overlying construction's growth is a modelling convenience, not a direct measurement.
• 2009 vintage. Reflects the 2008 SNA as newly adopted; land-value methodology has since advanced considerably, including hedonic approaches (Larson, 2015) and granular mass-appraisal projects (Davis, Larson, Oliner and Shui, 2019) that address land/structure separation with far more data than the ABS had in 2009.
• No LVT or tax-policy content. The paper makes no claim about land value taxation, incidence, or revenue; any Georgist inference is drawn by the wiki, not asserted by the authors.

Bears On

• Research: New Estimates of Value of Land of the United States — corroborates, from a different national statistical tradition, the same order-of-magnitude land share of national wealth, and documents the residual method's known weaknesses that Larson's later hedonic approach was built to address.
• Research: The Price of Residential Land for Counties, ZIP Codes, and Census Tracts in the United States — this paper's account of the land/structure separation problem is the historical backdrop against which the FHFA's later, more granular land-price project should be read.
• Problem: Most of the modern rise in the capital share is land, not capital — supports at one remove: the land/structure decomposition central to that page's claim is, as this paper shows, a genuine measurement exercise, not merely a stylized fact.

See Also

• Economic Rent
• New Estimates of Value of Land of the United States
• The Price of Residential Land for Counties, ZIP Codes, and Census Tracts in the United States
• Metropolitan Land Values
• Most of the modern rise in the capital share is land, not capital

Sources

1. Andrew Cadogan-Cowper and Peter Comisari (2009), "Recording Land in the National Balance Sheet," LG/15/11/1, Information Paper for the 15th Meeting of the London Group on Environmental Accounting, Wiesbaden, 30 November–4 December 2009. PDF via UN Statistics Division — used for all findings, figures, tables, and quotations; fetched and read in full (13 pages) this session.
2. United Nations, European Commission, IMF, OECD, World Bank, System of National Accounts 2008 — used via the paper's own quotation for the 2008 SNA's definition of land (para 10.175) and its treatment of location as a quality/volume factor (para 15.67); not independently re-fetched this session.