ABARES Farmland Price Indicator - Measurement Framework

ABARES Working Paper

Published 28 November 2024

Author: Will Chancellor

Introduction

This report presents the underlying measurement framework used for the ABARES Farmland Price Indicator , and is based on the working paper by Boult et al. (2023). It outlines the use of administrative data supplied by CoreLogic in a stratification model to estimate average Australian farmland prices. These averages are based on parcels of land transacted in each given period, representing actual market activity.

The previous iteration of these estimates, as described in Boult et al. (2023), applied weights using the land area of ABARES Australian Agricultural and Grazing Industries Survey (AAGIS) farm records, to derive estimates that represented the value of all farmland stock. This approach had some important advantages such as the potential to substitute land value data collected using the farm survey, and to provide an indication of farmland value for land parcels which would rarely be available for sale. However, the use of AAGIS farmland area weights added an additional layer of measurement complexity and made the estimates somewhat difficult to interpret. For example, in the context of residential real-estate prices, the average value of houses in a suburb would be represented as the entire population of houses in that suburb — rather than the houses sold in a given period. In addition, more detailed estimates were less stable (e.g. state level estimates), and could only be published as index numbers.

The revised approach now used in the ABARES Farmland Price Indicator assigns transaction records to strata according to the CoreLogic land area of the parcels transacted. This approach is intuitive and consistent with real estate price estimation, and is relatively simple to estimate, given that it does not require weights from a different dataset (AAGIS). The estimates reflect the market transactions and appear to be more stable, facilitating for the publication of ‘price per hectare’ estimates, rather than indexes. More detailed farmland price averages are also possible using this approach, including estimates by quarter, region, and farmland type. The use of administrative data means that there is no additional statistical survey burden on farmers, and there is a very large volume of data for robust analysis.

An estimation pipeline has now been developed whereby the raw data from CoreLogic is cleaned, processed, spatially linked and stratified, to generate a series of farmland price estimates. These statistics are packaged and presented as a free downloadable product – now available through the ABARES Farmland Price Indicator.