APD8500 - Special accounting schemes: Sampling

What is sampling?

Many special schemes involve statistical sampling. This means that the aircraft operator will be using information about a sample of passengers to calculate the amount of duty due on the rest. The results of the sample are used to calculate the percentages of passengers carried at each rate of duty and some or all exemptions from the duty.

Points to consider

Where statistical sampling is involved the CAT has two specific areas of responsibility. You should:

• advise the aircraft operator of the issues they should consider; and
• assess a proposal to determine whether a scheme is likely to be acceptable.

Before agreeing a sampling method, the CAT must ensure that any samples taken are representative. This means that all the following should be true:

• the sample(s) take account of any variations or changes within the aircraft operator’s business that might affect APD liability;
• there is no difference between the types of passenger in the samples and those not;
• sampling is repeated at suitable intervals to reflect any seasonal change within the aircraft operator’s business;
• the sample(s) are large enough to ensure an accurate estimate of liability (guidance is given below;
• the liability for the passengers within each sample is calculated correctly;
• any information used is accurate and a true reflection of the current pattern of business. (It is legitimate to use historic data but only if the business has not changed).

Note: Auditors are not expected to provide detailed statistical advice and should contact the CAT if technical assistance is required.

Procedure for approving a sampling method

The CAT should:

1. Consider each of the issues detailed below and ensure that the sampling method takes them into account.
   
2. Issue a letter of approval in the normal way – see APD8800.
   

What factors are there within the business that might affect the aircraft operator's liability to APD?

The aim of special scheme sampling is to estimate the percentages of persons carried at each rate of duty. For a sample to be accurate it is essential that the method takes account of any factor which might affect these percentages. Common examples include:

• origin (that is starting airport);
• destination;
• time of day;
• time of week;
• time of year; and
• special events (such as festivals and group bookings, if not covered by the time of year).

Action:

• Take all significant factors into account during sampling; and
• Consider how best to group flights together and how often sampling should take place.

How should flights be grouped together?

To take further account of the factors affecting liability, many schemes involve “flight grouping”. This involves dividing the aircraft operator’s flights into a number of groups so that all the flights within each group have, or are expected to have, similar percentages of passengers at each rate of duty. Separate samples are then taken for each group.

Example 1:

A aircraft operator flies from Glasgow and Heathrow to the Mediterranean throughout the year. An analysis of data from last year highlighted a marked difference in connection patterns between passengers at Glasgow and those in London. A study of past trends suggests that a quarterly sample would take reasonable account of seasonality. The aircraft operator decides to group by airport of departure and by quarter. He has eight groups (4 quarters times 2 airports) and so takes eight samples during the year.

Example 2:

A scheduled airline operates out of London to Belfast and Dublin. They do not have much seasonal variation beyond the winter/summer divide, but morning flights have more onward connections than average and Belfast more business class passengers. The airline decides to group according to destination (because of the differing split by class), by time of day because of the onward connections and by season (winter or summer). They also have 8 samples altogether.
Action:

• Take the main factors such as the airports of departure and destination into account;
• Consider any further issues, such as time of week, when the sample is selected.

How often should samples be taken?

If the time of year affects the types of passengers carried, and hence the APD due, then samples should be taken at intervals throughout the year. Further information on frequency of samples including an example of sampling frequency can be found in Notice 551 SpecialAccounting Schemes for Air Passenger Duty.

Action

Discuss with the operator which sampling frequency most closely mirrors the seasonality in their own business.

How should the samples be selected?

There are three possible methods for selecting the flights (or passengers) which will form each sample. The operator can:

• select flights (or passengers) at random;
• select all flights for a specified period (usually one week); or
• identify further influential factors that need to be taken into account and ensure that the sample reflects this. For example, if 30% of passengers travel on morning flights then the aircraft operator might ensure that 30% of the sample was also taken from morning flights. The sample becomes more representative and thus more accurate.

Action

• Check that the operator will select the samples as fairly as possible.
• Ask whether the types of passengers selected for the samples are likely to be any different from the rest.
• If the answer is yes then contact KAI for advice because the method may need to be modified.
• Ask for as much information as possible about seasonality, routes and historical passenger numbers so that you can judge whether the sampling method is likely to give a fair and representative picture.
• Whichever method is chosen, ensure that the operator includes full details of how they will select the sample within the special scheme proposal.

How big should the samples be?

It is important to take a sample of sufficient size. A sample that is too small may be inaccurate. The correct size depends on a number of factors and should not be agreed without statistical advice. However, the following table gives an indication:

If the proposed sample size meets these guidelines then the method will be acceptable.

If the sample size does not meet these guidelines then the sample may still be acceptable, but you should consult KAI for advice.

If these guidelines are met then the sample results will be within +/-3% of the true APD liability, 95% of the time.

Action

• For each flight group, consider the total number of passengers for which the sample will be used. For example, a group of all Heathrow flights to America in the summer season might contain 50,000 passengers.
• Use the table to give a guide, but no more than that, to the appropriate sample size (2,500 passengers in the case of the above example).

Common problems

Is the sampling method acceptable?

If the proposed method takes the variations within the aircraft operator’s business into account and the sample size falls within the recommended range then it may be assumed that the method will be broadly acceptable.

What happens if a sample is disrupted (for example by industrial action)?

The CAT will consider all factors and advise a suitable way forward.

For further advice please contact the CAT.