Regular revision process

Each publication contains three sets of estimates:

• initial estimates for the most recent calendar year – which are based heavily on proxy estimates and projections
• second estimates for the previous calendar year – which will still be based on incomplete data
• final estimates for earlier calendar years – which are based on complete and final data

Many of the data used in calculating income from farming for a given year will only become available during the following year, and some will only be available two years after the year of interest.

In particular, TIFF estimates rely on the final results of the Farm Business Survey. Farm business survey estimates for a given financial year can only be collected after the completion of the financial year, and data collection and processing takes a future year. This means the estimates are not available until the second year after the year of interest.

This means that initial TIFF estimates for the most recent calendar year will contain a large number of forecasts, often based on projecting past trends.

When second estimates are released in the following year, this means they can differ substantially from initial estimates.

The difference between initial and second estimates of TIFF can be particularly substantial because TIFF is a profit margin calculated as income minus costs.

Income and costs are similarly-sized large figures, giving a relatively small difference as the value of TIFF. This means that a small percentage update in estimates for income or costs may automatically lead to a large percentage revision in TIFF. For example, if income = 100 and costs = 96, then TIFF = 4. If we then update the estimate for income upwards by just one per cent, and for costs down by just one per cent, income = 101 and costs = 95, so TIFF = 6, a 50 per cent increase in the value of TIFF. So reasonably small updates in the estimates of income and costs can lead to what may seem an substantial 50 per cent change in TIFF.

Final estimates will generally differ much less from second estimates, as the second estimates will contain a higher proportion of final data.  

Any methodological changes adopted are applied to historical years, back as far as deemed necessary, at the time of publication.  No further revisions are made until the next account is prepared.

Any errors in the account calculations will normally be corrected at the time of the next account being published, unless the error is considered large enough to warrant a more immediate correction.