JCPSG | TRAC Guidance

Part III - Annual TRAC

Section D: All other costs

1. All costs should be directly attributed, or attributed via cost drivers, to activities and to academic departments.
2. The techniques used are shown on the pyramid of costs

   Cost attribution to activity
   Time allocation to activity

3. Most research staff salaries are directly charged to Research. Academic staff costs (and those of other researchers) are attributed to activities using academic staff time. This was covered above. This Section covers:
   • the subsequent attribution of academic staff time on Support to activities;
   • the direct attribution of some non-staff costs in academic departments to activities;
   • the attribution of all other costs, mainly Support staff in academic departments, most non-staff costs in academic departments, and the costs of central service departments, to activities.
4. Costs should be attributed to:
   • academic departments (and then grouped into discipline type)
   • activities (T, R, O, O(CS), S);

     and then

   • from O(CS) to T, R, O, S
   • from S to T, R, O
   • finally, to lower-level activities (PF, NPF, research sponsor type).
5. This is shown in the figure: Overview of the cost attribution process.
6. Costs are either attributed directly (where they relate specifically to one department or activity), or cost drivers such as space, or time are used.
7. Costs should be directly attributed to activities where possible. Some costs have already been directly charged through their internal coding (e.g. directly incurred costs on research grants and contracts). Other activity-specific non-staff costs in academic departments could be coded at source to assist with this. This would include Teaching placement costs, undergraduate student fieldwork, consumables used for institution-/own-funded Research or PGRs. Whether they are coded at source, or subsequently attributed using a cost driver, these costs should be classified as a direct cost of their activity. They should not be classified as a Support cost.
   
   
   An illustration of how cost drivers can be used is given in this table.

Chapter D.1 Cost pools and cost drivers

1. It is efficient first to group costs before their attribution. These cost groups are often called cost pools.

Cost pools

2. Costs in any one cost pool are attributed using the same cost driver. A cost pool may relate to an activity, or a Support cost:
   • activity cost pools represent the costs of activities. There are three main activity cost pools – one for each of Teaching, Research, and Other. Additional cost pools are needed for sub-activities (PFT, NPF T; each research sponsor type; O and O(CS));
   • Support cost pools are intermediate holding places. The costs of a finance department may be separated into a number of cost pools – relating to payroll, fees, all other areas. A different cost driver is used for each cost pool (staff time/numbers, student FTEs, all other costs) when attributing its costs to an activity cost pool.
3. All cost pools need to hold information by academic department; and need to differentiate between Support costs and direct costs of the main activities of Teaching, Research, and Other (to allow the calculation of the indirect cost rates).

Cost drivers

4. Cost drivers are used at various stages of the cost attribution process. Cost drivers are used for those costs that cannot be directly allocated.
5. Costs attributed to Support cost pools use cost drivers to attribute their costs to activity cost pools. For example, estates costs would be assigned to an estates cost pool and would be attributed to academic departments and activities on the basis of space used.
6. Some costs do not need to be routed via a Support cost pool, as they can be directly attributed to an activity cost pool:
   • catering, conferences, residences – to Other
   • trading companies – to Other (or Research or Teaching, depending on the activity being carried out)
   • research support department – to Research.
7. However, cost drivers will still be applied to costs in the last cost pool to attribute those costs to academic departments, and to research sponsor type.
8. A wide set of definitions of cost drivers can be used:
   • a measure of the quantity of resources consumed by an activity
   • those factors responsible for variation in the cost of an activity
   • a measure of the benefits received
   • a measure of how often activities are performed and the effort involved in carrying them out (an activity driver – e.g. the number of students or staff).
9. Cost drivers used for attribution to departments are not generally the same as those used for attribution to activities.
10. A wide range of cost drivers could be used. Some are used by some institutions in their existing resource allocation models (RAMs). However, RAMs have historically had a different focus and it is unlikely that all cost drivers currently used in RAMs are relevant:
    • TRAC requires an attribution to activities (few RAMs have required this);
    • the data used in the RAM may not be up-to-date and may be drawing on planned rather than historical data;
    • there should be no ‘motivational’ element in the cost drivers for TRAC – this can be important part of a RAM;
    • the cost drivers in the RAM are often many and detailed and their use within the extra dimensions of TRAC (activities, PF and NPF, research sponsor type) may prove very complex.
11. It is important to avoid confusion over the two sets of drivers, and the information that results. It is good practice for the differences to be minimised, and clearly understood. Many institutions are planning to bring their RAM and TRAC methods in line as they implement TRAC fEC and allocate the additional research funding that they are receiving – see Annex 5.
12. Sensitivity analysis can be carried out to inform management of the impact on different cost drivers on key cost pools.
13. TRAC does not require a large number of cost drivers. Institutions have considerable discretion to choose which cost drivers to use and at what level of detail. Some areas to be considered are:
    • the costs that are already directly charged to departments in institutions’ management accounts (for example, equipment costs, space costs, in some institutions);
    • the availability of information held for other purposes (for example, information on the use of library and learning resources may be recorded for use in resource allocation and strategic planning in some institutions);
    • the cost of collecting the information (for example, comprehensive information of the use of IT facilities may be costly to collect).
14. Some common cost drivers that might be suitable to use are:
    • headcount and/or academic/research FTEs – sometimes including clerical, administrative and technical staff;
    • headcount and/or student FTEs, distinguishing between PGR, under-graduate and post-graduate taught students;
    • square metres of usable space occupied by departments;
    • total Research costs;
    • total non-pay costs;
    • academic or total staff time;
    • all DI costs.
15. An illustration of how these cost drivers can be used is given in this table.
16. Academic staff time should only be used as a cost driver if there is no other better alternative. When used, it should include the time directly attributed (by Research Assistants) as well as that attributed on time allocation schedules (by academic staff).
17. Income is unlikely to be a robust cost driver as it is not a good indication of the use of resources by departments or activities.
18. If direct costs are being considered as a cost driver, then two factors should be considered:
    • the purchase cost of large items of equipment should not be included;
    • attribution to Other activities will be lower than it would be if academic staff time was the driver. Direct staff costs may therefore form a more appropriate driver than direct costs.
19. In some cases a combination of cost drivers may be appropriate e.g. the use of both staff and student numbers to attribute some types of library costs.
20. In some cases it may be appropriate to weight elements of the cost driver/s used e.g. Arts students may receive a weighting of 2, compared to 1 for Engineering students, in attributing costs to academic departments for the use of the library. These weightings should be evidence-based (e.g. the number of library loans) where appropriate – see Annex 11.
21. The use of the cost drivers at the ‘broad-brush’ level shown in the following table are acceptable for TRAC, provided the institution considers it a fair and reasonable representation of the use of the resources being attributed. The exception is for estates and library/learning resource centre cost drivers, which should be reviewed in more detail.
22. All cost driver data should be robust.
23. The table shows cost drivers that could be used to attribute:
    • costs to academic departments (column 2);
    • costs to activities – Teaching, Research, Other, Other (Clinical Services), and Support (column 3);
    • the costs of Support activities to Teaching, Research and Other (column 4).
24. This table does not show the next level of attribution – Research costs to research sponsor type, Teaching costs between PF or NPF, or O(CS) to Teaching, Research and Other. These are discussed in Chapter D.3.
25. Costs can be attributed at a high level. Column (1) in the table above refers to high-level accounts codes, aggregated as required in HESA FSR Table 6 (accessible via Annex 1). These high-level codes can be used in TRAC (along with department-level codes for attributing costs to departments. This means that only a few high-level cost drivers are required.
26. A cost driver does not need to be separately identified for every account code – this would lead to an extremely onerous attribution process. The lower the level of code, the more detail needs to be extracted from the accounts – and simple spreadsheet systems might need to give way to more complex databases with computerised data transfers. Cost pools are used to group costs and simplify processes.
27. It is good practice, however, to consider the use of more detail than that shown in the table, as it leads to better quality cost attribution. It may also be useful if an institutions wishes to meet other cost objectives (e.g. costing to lower-level activities not required under TRAC). In particular, an institution might wish or need to analyse their accounts codes at a lower level than shown here, e.g. estates costs into building cost pools, or library into periodicals/staff/books etc. This allows them to use many more cost drivers than illustrated above. This also meets TRAC requirements, but is not necessary for TRAC, and of course involves more work.
28. Others may consider that more complex attributions are necessary properly to represent the use of given resources, e.g. finance department split into payroll (using a cost driver of staff numbers); accounts payable (total non-pay costs); accounts receivable (total income); etc. Materiality is relevant here.

Chapter D.2 Support costs

1. Support costs for each activity and for each department should be identified and totalled. This will consist of the totals of all the Support cost pools, prior to their attribution to activities. It provides the totals that will be included in the indirect cost rate and estates charge calculations.
2. Support costs are costs that are shared, and cannot be directly attributed to activities. They include indirect costs and estates costs. They are defined in Annex 6.
3. Estates costs of central service departments are initially held in the estates cost pool, but are then attributed to the cost pools for central service departments. This element of estates is therefore attributed to activities along with other central service costs, and they are included in the indirect cost rate, not the estates charges.
4. Some of the costs of estates, laboratory technicians, and major research facilities will be recorded in the financial accounts (on the project cost records) as Research. These are historically based estimates and should not form the basis for cost attribution to annual TRAC. Actual costs should be attributed to activities using cost drivers.
5. The Support cost total, together with directly incurred costs and other costs that have been charged directly to an activity, should agree with the financial statements including exceptional costs, less the net cost of capital employed and net infrastructure adjustment.
6. Support costs are to be attributed to Teaching, Research and Other, as shown in the table above.
7. In some cases it will be a straightforward process (e.g. research administration to Research; most of Registry to Teaching – although some costs will need to be attributed to Research for the PGR student usage of Registry).
8. Otherwise, a reasonable cost driver might be all ‘other costs already attributed’.
9. Non-staff costs (and support staff) in academic departments which are not directly chargeable to an activity should not automatically be attributed to activities on the basis of academic staff time, nor should it be assumed that they only relate to Teaching, or only to Research. An example of a robust approach is periodically to provide heads of department or equivalent with a list of these costs and ask them to indicate the appropriate percentage split between Teaching, Research and Other.

Chapter D.3 Further levels of attribution

1. Having attributed costs to departments and activities, and from Support to Teaching, Research and Other, costs must then be attributed:
   • from O(CS) to T, R, O, S
   • from T to PF T and NPF T
   • from R to research sponsor type.

Clinical services costs to T, R, O, S

2. Some staff and non-staff costs (in medical or dental schools) will be identified as services provided to the NHS by the HEI under knock-for-knock. These should be attributed firstly to a ‘clinical services’ cost pool (a lower-level O activity – O(CS)); then reattributed to T, R, O or S on the basis of the services being provided to the HEI by the NHS. This is covered in Annex 10.

Cost drivers to attribute T to PF and NPF; and R to research sponsor type

3. T costs should be attributed to PF and NPF; and R to the seven types of research sponsor. The seven types of research sponsor will then be aggregated to provide the totals of PF R and NPF R.
4. Some non-staff costs, as well as staff costs, will already be identified by research sponsor type as they are recorded in the financial accounts as a directly incurred cost on a research grant or contract.
5. Alternative methods of attributing T time to PF T and NPF T, and R time to research sponsor type were suggested in the context of attributing academic staff time. These use either staff time (via the time allocation schedules) or a proxy such as the number of Research Assistants working on projects in each sponsor group, and PGR student numbers. These attributions could then be used as the basis to attribute all non-staff costs. In non-laboratory departments the heads of department need to give an informed allocation, as RA and PGR numbers are unlikely to be a useful proxy there.
6. Institutions should ensure that the methods they are using to attribute Research costs to research sponsor type are robust. This will involve considering the use of:
   • time, collected via the time allocation schedules; or
   • a proxy: such as the numbers of Research Assistants and PGR students; or total direct costs (i.e. excluding overheads);
   • a weighted proxy (a Research Assistant weighted against a PGR student).
7. When preparing their costs for the annual TRAC return to the Funding Councils, institutions should ensure that the total of the Research indirect costs that has been attributed to the PGR activity category is the same as that that could be calculated using FTE weighting of 0.2 for a research student (compared to a weighting of 1.0 for an academic and research staff FTE working on Research). This is irrespective of the cost drivers that have been used in their cost allocation model when estates costs are allocated to Research.
8. Institutions should review their cost drivers for PGR costs so that they are in a position to inform a UK study of PGR weightings in 2006.
9. Chapters D.4 and D.5 provide more detail on the requirements for attributing library and estates costs. Annex 6 (definitions) provides lists of costs within each activity type.
10. Annex 14 provides guidance on attributing costs in the following specific areas:
    • national museums and libraries
    • sports facilities
    • parks and gardens
    • botanical gardens
    • farms and woodlands
    • veterinary schools
    • sales and leasehold schemes
    • costs covered by special initiative funds.

Chapter D.4 Libraries

1. The costs of libraries and learning resource centres should be attributed on a robust basis. Institutions should:
   • allocate costs directly to an activity or academic department where possible, and otherwise, use appropriate proxies;
   • review the results of this attribution with the librarian, and obtain their agreement as to the reasonableness of the overall allocation.
2. It is good practice to also:
   • review the cost-benefit of collecting more information to inform attribution (data could be collected through data samples, occasional reviews, and/or measurement);
   • consider analysing further cost elements into different cost pools, and using different cost drivers to attribute them. For example, periodicals, specialist collections, staff by area of responsibility.
3. The minimum requirements are given in Annex 11 which should be followed. More complex methods are also described but these remain optional. Materiality should be considered.

Chapter D.5 Estates

1. The TRAC Support costs include estates costs (or premises costs).
2. They are defined consistently for the annual TRAC process and for calculating the estates charge-out rates. They are:
   • premises expenditure (as defined in the HESA FSR – accessible from Annex 1);
   • the TRAC infrastructure adjustment; and
   • initially, laboratory technicians and all equipment and research facilities (these are then excluded from the estates costs used to calculate the estates charge-out rates).
3. They include all the costs of all support staff that relate to these areas. The cost of capital employed should not be included (it is an indirect cost).
4. Further detail of what is and is not included in estates costs is given in Part IV .
5. When attributing estates costs in the annual TRAC process, the estates costs of residences should be attributed to Other.
6. Non-residential estates costs should be attributed to central departments and to academic departments.
7. Estates costs of academic departments should be attributed to Teaching, Research and Other.
8. All attributions should be based on the use of that space.
9. No later than August 2007, the use of this space should take into account at least four different types of space. A weighted cost per square metre should be used to attribute the space used.
10. The Research estates costs should be attributed to research sponsor type; and the Teaching estates costs between PF and NPF Teaching.
11. The estates costs of central departments should be identified, and assigned to the central service department cost pools (and included in the indirect cost rates). Where central service functions such as libraries, registries, etc, are significantly devolved to academic departments, the estates costs relating to these areas could either be included in the estates total or be included in the indirect cost total.
12. Robust methods should be used to attribute shared space (e.g. laboratories used for both Research and Teaching, desk space used for all of Other, Research and Teaching activities, shared seminar rooms) both to academic departments and to activities. Shared space should be attributed on a reasonable estimate of the proportion of use of that space, not on the basis of ‘predominate use’, or a proxy based on other dedicated space allocations. Office space could be attributed using academic staff time data.
13. The use of space by PGR students should be separately identified on a fair and reasonable basis as part of this.
14. When preparing their costs for the annual TRAC return to the funding councils, institutions should ensure that the total Research estates cost that have been attributed to the PGR activity category is the same as that that could be calculated using FTE weightings of 0.5 and 0.8 for a research student in non-laboratory and laboratory departments respectively (compared to a weighting of 1.0 for an academic and research staff FTE working on Research). This is irrespective of the cost drivers that have been used in the cost allocation model when estates costs are allocated to Research.
15. Estates costs for PGR students should not be extracted separately from the estates costs of other Research activities before estates charge-out rates are are calculated.
16. Institutions should review their cost drivers for estates costs so that they are in a position to inform a UK study of PGR weightings in 2006.
17. The requirement to identify four types of space and calculate a different cost weighting for each is to ensure that departments are given the right total costs for their mix of space. Weighted space costs provide institutions with a better method of attributing space costs to departments, and between T, R, and O in those departments.
18. The four different space categories might be :
    • low-cost space (e.g. barn, shed, some animal houses)
    • desk or office
    • serviced laboratory or facility (e.g. laboratory, computer laboratory)
    • highly serviced laboratory or space (e.g. laboratory with air-conditioning, or a major facility).
19. If an institution does not have one of these types of space, they can identify only three types of space.
20. The £ per sq metre cost for each category should be calculated for benchmarking purposes – see Part II. The differentials would recognise the respective costs of building, maintaining and servicing each category of space; and their respective utilisation taking into account use by all activities.
21. A step by step approach to the attribution of estates costs is available in Annex 12 . A worked example is available in Annex 13.

Chapter D.6 Steps in cost attribution

1. The cost attribution process consists of several steps:

   set up system collect cost driver data extract and reconcile costs attribute to activities and departments calculate results test figures for reasonableness report review methods for reasonableness and amend, as appropriate

i. set up system

2. A spreadsheet or database system can be used. This would cover the following:

   Costs for attribution

   Containing:
   • list of cost totals from consolidated financial statements
   • calculation and addition of cost adjustments – see Annex 15
   • analysis into high-level accounts code groups.

   Time allocation data (note: time is a cost driver)

   Containing information from the time allocation process which shows the percentage split of time for each department between:
   • T, institution-/own-funded R, all other R, O, O(CS) and S
   • if collected, PF T, NPF T, research sponsor type, and sub-categories of S.

   Other cost drivers

   All other cost drivers would be included by department, and by activity, as appropriate, e.g.:
   • space (by department, and by activity);
   • staff numbers by department (FTE and headcount);
   • student numbers by department (FTE and headcount – split PGRs and all other);
   • research costs by department;
   • % allocations given by heads of department (e.g. for libraries, support staff, non-staff costs in their departments, external Research by research sponsor type if required);
   • numbers of Research Assistants and PGR students, by research sponsor type;
   • percentages to be used to allocate O(CS) to T, R, O and S – see Annex 10.

   Attribution to academic departments

   Academic department costs would have been attributed automatically to academic departments through their accounts codes.

   Central department groups would be attributed either:

   • automatically to O – i.e. for residences, and catering and conferences; much of Other Services Rendered and trading companies; recoverable/reimbursable salaries (medical and dental schools)
   • through application of the cost drivers – see table.

   Attribution to activities

   Most costs would be attributed to activities on the basis of their identified cost drivers.

   Some costs would have been identified as S as they were being attributed to T, R and O.

   Some would be directly attributed to activities – e.g. research grants and contracts to R; the ‘O departments’ to O; general educational expenditure to T.

   O(CS) would be attributed to T, R, O, and S on the basis of cost drivers.

   Attribution to sub-activities

   T to PF and NPF – on the basis of time (if recorded) or student numbers.

   R to research sponsor type on the basis of time (institution-/own-funded, and research sponsor type if available); and/or on the number of Research Assistants and PGR students; or (for non-laboratory departments) head of department estimates.

   Income

   As income is to be reported alongside costs in the annual TRAC return – see Annex 16 – this would usefully be included as part of the cost attribution model.

   Management information

   The ‘five figures’ for reporting would be calculated (along with the income).

   Surpluses/deficits would be calculated by activity, and, for internal information, by department as well as average T costs per student FTE; and so on.

3. Normal audit trails should be included as part of the system design to ensure data integrity, and it is good practice (periodically) to carry out an independent review of the logic etc. As part of this a running check on the totals being allocated should be done (to ensure they remain reconciled with the consolidated financial statements including exceptional items before taxation plus cost adjustments).
4. TRAC quality assurance requirements, including information on maintaining audit trails, are given in Part II.

ii. collect cost driver data

5. Cost driver data should be robust – selected and applied according to TRAC principles and conventions and recorded with appropriate audit trails.

iii. extract and reconcile costs

6. Costs should be extracted from the nominal ledger (which reconciles to the audited financial statements), and from the cost adjustments – see Annex 15.
7. Costs should be total expenditure plus exceptional items before taxation.
8. Where subsidiary companies are consolidated within institutions’ accounts their costs should be reported as part of TRAC costs. These should be attributed to T, R, O or S, on the basis of the activity being carried out in the company.
9. It is good practice for data to be collected at as high a level as possible to avoid complex data manipulation. Where data are not available at the required level (i.e. available only at an aggregate level, or at a very detailed level by accounts code) then institutions should consider:
   • using proxies (e.g. notional staff salaries to extract staff costs that require a specific attribution – e.g. 100% to R – from a general staff cost pool. The balance of costs in this pool will be attributed using the data from the time allocation schedules);
   • using data from other accounting systems – e.g. the management accounts (these might not be reconciled at all levels, but would provide, for example, a reasonable indication of the split of some costs between departments);
   • drawing from information provided for other external reporting e.g. HESA Table 6 data – accessible via Annex 1. Institutions have found the internal working papers to arrive at these HESA figures can be of use in informing their departmental split of costs, where otherwise lower-level accounts codes would need to be allocated.
10. Significant internal recharges and large adjustments in departmental costs (for example) would need to be examined to identify their purpose and relevance, and their impact on TRAC. Institutions have found that these can be significant, and unravelling them can be complex.
11. The costs would be entered into the cost allocation model and the total/s reconciled to the audited financial statements (plus adjustments). Reconciliations from cost attribution models involving significant use of management accounting information, to the published financial statements can be complex. Large cost differences can arise as accounting standards are fully applied to year-end figures, and these differences need to be identified and attributed carefully. A full audit trail should be maintained.

iv. attribute to activities and to departments

12. As shown above, costs need to be attributed to:
    • departments
    • activities
    • lower-level activities.
13. All costs should be attributed in full to all activities – no activity should be costed on a marginal basis.
14. Costs can be attributed using step-down or cascading - and this can easily become quite complex (with step-ups, and numerous iterations). A simple example here is the attribution of estates costs to a finance (administration) cost pool; and then finance to academic departments. Alternatively, finance costs could be attributed to the estates cost pool. The JCPSG publication, Costing Guidelines (page 55) – accessible via Annex 1 - illustrates this.
15. Step-down attribution is not necessary for TRAC, but could be useful if institutions wish to establish the costs of a service department to enable comparison with commercial providers or to inform pricing (for example). Use of step-down attribution might materially alter the costs attributed to activities. It is for individual institutions to determine whether step-down attribution or cascading is appropriate or necessary.

v. calculate results

vi. test for reasonableness. See Part II

vii. report

viii. review methods for reasonableness and amend as appropriate.