At the Farm Structure Working Group in October 2017 Eurostat proposed that the utilised agricultural area from the common land units could be allocated to the agricultural holdings and subsequently these common land units would be removed from data dissemination.

The recommended method suggests an allocation done proportionally on the basis of the grazing livestock at the lowest geographical level available in the records (for both common land units and agricultural holdings). This would allow that:

• The common land area will be included in area totals and breakdowns.
• The common land units will not be counted in the number of holdings as they are artificial units.
• The average size of holdings and other indicators will have reduced bias.

NSI would be better placed than Eurostat to do the allocation. They might have better information regarding which agricultural holdings have sent livestock to graze on common land in the same geographical area. In the absence of this information, countries or Eurostat can do the allocation to all agricultural holdings with cattle, sheep, goats and equidae in the geographical area of the common land unit.

Below you will find an example on how to allocate common land recorded under common land units to the relevant agricultural holdings, proportionally on the basis of the livestock (bovine, sheep, goats, swine and poultry) in the case of a sample survey. The allocation in case of a sample survey is less intuitive than in the case of a census because of the need to use weights.

The example also illustrates the change in the average area when the proposal is implemented as compared to the situation when common land units are kept in the records.

Statistical allocation

The example is given for a very reduced number of common land units and agricultural holdings and for only one variable - utilised agricultural area (UAA), but the procedure is the same for large numbers of common land units and agricultural holdings and for the other variables.

Step 1

Identify the lowest geographical level available in the records for which user agricultural holdings can send livestock to graze on common land.

Step 2

Group the records by the identified geographical level.

Step 3

Calculate the weighed UAA that is going to be allocated to the agricultural holdings of the same group geographical area.

The example table below considers a group where there are 3 common land units (calculated values are presented in green).

Step 4

Distribute the UAA among the agricultural holdings of the same group/geographical area that have at least one grazing animal.

The example table below shows the data for a hypothetical case, where 50 agricultural holdings are in such conditions.

The following table shows how to calculate the adjusted values of UAA corresponding to the agricultural holdings after addition of the common land from common land units.

Analysis of the overall average areas in different scenarios

The table below shows additional data used to compute the averages.

Scenario 1 – Average UAA considering common land units and their land

Scenario 2 – Average UAA disregarding common land units and their land

Scenario 3 – Average UAA after allocation of common land from common land units and removal of common land units

It results that by implementing the proposal of allocating the common land from common land units to agricultural holdings and removing the common land units, the average UAA increases from 33.1 hectares to 34.3 hectares. This increase is not significant. The increase can be significant for the breakdowns, as illustrated in the following two sub-sections.

On the other hand, if common land units are simply removed from the records without doing the allocation, the average UAA is 18.6 hectares.

Analysis of the average areas by age of manager in different scenarios

Consider that the average areas are disseminated by age of manager and that the first 8 agricultural holdings (holding IDs from 4 to 11) have a manager whose age falls in the same age band.

Common land units should not have recorded information on managers. Thus, the average UAA under scenario 1 cannot be computed because it is unclear how to distribute the number and area of common land units by age bands.

For scenario 2 and scenario 3, the same computations as above but now only for the first age band lead to the following results.

Scenario 2 – Average UAA disregarding common land units and their land

Scenario 3 – Average UAA after allocation of common land from common land units and removal of common land units

Analysis of the average areas by size of agricultural area in different scenarios

Consider that the average areas are disseminated by size of agricultural area. 6 agricultural holdings have the utilised agricultural area of 100 hectares and over.

The same computations as above but now only for the largest size group lead to the following results.

Scenario 1 – Average UAA considering common land units and their land

Scenario 2 – Average UAA disregarding common land units and their land

Scenario 3 – Average UAA after allocation of common land from common land units and removal of common land units

Conclusion

There are two possible ways of disseminating the UAA averages, both of which being correct as long as the meaning of the averages are communicated to the user:

• The average area over agricultural holdings, disregarding common land units and their land (scenario 2 in the given example)
• The average area over agricultural holdings, after allocation of common land from common land units and removal of common land units (scenario 3 in the given example).

When using scenario 1, the bias is not likely to be significant when the averages are computed over the whole population but can be significant when the averages are computed over population breakdowns.