Grain production and value: experimental analysis using administrative data

The Australian Bureau of Statistics (ABS) is working to modernise the way we produce official agriculture statistics. We are working with agriculture experts to co-design methods that maximise the use of existing administrative data. As well as improving the timeliness and regional detail of statistics, this approach may reduce reporting burden for farmers and the agriculture industry.

A significant existing administrative data source is the levy payer data generated during the sale of agricultural commodities. Levies are collected by the Department of Agriculture, Water and the Environment (DAWE) for the purpose of funding agricultural research.

ABS’ initial investigation of the levy payer data shows that it has the potential to replace the need to collect certain information from farmers via surveys in the future. This analysis considers the suitability of levy payer data as an important component in producing national and state level statistics for a selection of high value broadacre crops, specifically: wheat, oats, barley, sorghum and canola.

Further work will be required to compare the levy payer data with survey-based data over additional time periods and for smaller regions to determine its full utility. Investigation into additional data sources and methods for measuring the on-farm use of some grains will be required to complement the levy payer data in measuring production.

Levies are collected on the sale of agricultural commodities for the purpose of funding agricultural Research and Development Corporations. The Levy Payer Register collects detailed information about levies paid on broadacre crops from individual producers. This information is available from the 2019-20 financial year and includes:

• Quantity of commodity sold by levy payer
• Amount paid to the levy payer for the sale of the commodity
• Business address of each levy payer (from which it is possible to determine an approximate farm location)

Using the existing levy payer data to produce a range of agricultural statistics has the potential to produce more timely statistics with greater regional detail and reduce the reporting burden on farmers by not asking them for information that is already held by government.

This analysis compares levy payer data with ABS data at State and National levels.

The ABS currently measures the contribution of agriculture to the Australian economy through an annual estimate of the Value of Agricultural Commodities Produced (VACP).

The VACP measures both Gross Value and Local Value by multiplying production by gross unit value or local unit value for each individual commodity type. Surveys are the main data source used in the calculation of these statistics.

Gross unit value

Gross unit value represents the price per unit of the commodity when it is relinquished by the agricultural industry. Gross unit value includes sales costs which are the marketing and transport costs required to move the commodity from the farm to a point of sale such as a fresh food market, a factory or for export.

Local unit value

Local unit value represents the price per unit that the farmer receives for producing the commodity and does not include the sales costs.

Production

Production of different commodity types is estimated from the annual agricultural survey or census and published in Agricultural Commodities, Australia.

Because levy payer data reflects the value paid directly to the farmer for producing the crop, the comparison with ABS data is at the local unit value. This is the dollar amount per tonne of grain sold.

At the national level, the levy payer local unit value varies between 1% less for canola and 11% less for barley than the VACP value. Wheat, oats and sorghum are 7% less than the VACP national local unit value.

At the state level there is greater variation between the two data sources with the levy payer value for oats in Queensland 27% higher than the VACP value, while wheat in Western Australia is 10% less than the VACP value. The local unit value of sorghum for both data sources is within 1% in both New South Wales and Western Australia. 

These differences between the levy payer and VACP local unit values are reflective of the scope and methods of calculation of the two data sources. One notable difference is that the VACP local unit value is derived by subtracting sales costs from the gross unit value.  This means that the sales costs impact the comparison between local unit prices. The consistently lower national levy payer local unit values may result from ABS’ current estimates of sales costs.

To compare production for the different broadacre crops, the quantity of commodity sold by the levy payer was compared to the quantity of commodity produced on-farm from the 2019-20 financial year ABS survey derived statistics.

At the national level in 2019-20, the levy payer and ABS production data are within 1.5% of each other for both wheat and canola. The levy payer production data for barley was 5% less than the ABS survey data, and for sorghum the levies quantity is 10% more than the ABS source. These levy payer quantities are all within the 95% confidence range for the ABS survey estimates. This analysis suggests that the levy payer data has the potential to be used to produce wheat, canola, barley and sorghum production statistics. 

The quantity of oats produced in the levy payer data is 52% less than in the ABS survey data. This is likely to reflect that a large proportion of oats are used on-farm for purposes such as stock feed. Because these oats are not sold, they are not included in the levy payer data. For crops where a significant proportion of the crop is used on-farm, additional data sources will be required to help estimate a production value using the levy payer data.

Two factors that could impact the accuracy of the levy payer data in estimating production are:

• Administrative errors in the quantity component of the levy payer data. ABS edit the data to adjust for inconsistencies between quantity and value information in the levy payer data to minimise the impact of this.
• The timing of crop sales. Broadacre crops are typically sold within the year after harvest, however, some businesses may store crops for longer periods before selling them and this may change from year to year.

This comparative analysis is the first stage in assessing the levy payer data as an important component for producing agriculture statistics. Future steps to further understand differences between the levy payer data and ABS survey-based estimates, and develop approaches for using the levy payer data are set out below.

Comparisons with other seasons 

The 2019-20 data is the first year that the levy payer data is available for analysis. Undertaking similar analysis with 2020-21 data will help to determine the consistency of the levy payer data in relation to ABS survey estimates.

Comparison at smaller geographic regions

Because the levy payer data records sales from specific locations, this data has the potential to support more detailed regional outputs than the state comparisons shown here. Future work could use the levy payer data to create Statistical Area Level 4 (SA4) production estimates. The levy payer data could also be used together with satellite derived crop mapping to produce more detailed regional statistics at the Statistical Area Level 2 (SA2) as has been done for sugarcane.

Methods for determining on-farm use

Because levy payer data only reflects crops that are sold, further work will be needed to investigate other data sources which could complement the levy payer data to understand the amount of on-farm use for certain crop types, such as oats. This will be important to fully understand production for all broadacre crops and would also provide a more detailed understanding of how crops are used after harvest.

Investigation of sales costs

A possible difference between ABS local unit values and the levy payer local unit value is the ABS’ estimate of sales costs. These reflect the amount spent on transport and marketing. Reviewing the way sales costs are calculated would be an important step in using the levy payer data as the basis for estimating local and gross value.

This initial analysis has shown that the levy payer data can be a valuable data source for producing broadacre crop statistics. This is the first stage of producing timely, regional broadacre crop statistics using a combination of data sources that reduce the need to survey farmers. 

The ABS will continue to investigate ways to use levy payer data to produce official agriculture statistics by producing experimental estimates for a range of agricultural commodities.

If you are interested in being part of the conversation about modernising the production of official agriculture statistics, please get in touch with us via email: agriculture.statistics@abs.gov.au by 31 October 2021.

07/09/21 A minor data amendment was made to the state data for oats to align with existing publications. This did not impact the analysis or findings.