Charging explanation

This document summarises the findings of the JCEI Charging Working Group.

1.Why not just meter usage?

Billing directly for measured usage would be complex, particularly given that the individual units of usage would be very small. The implementation effort for the datacentres would be considerable, and has not yet been assessed. And such a method of charging would create budgeting uncertainty within institutions, and might lead some to ration usage. So direct usage billing has been rejected for the moment, though it might be reconsidered at some point in the future.

2. So what else can be an indicator of usage?

Given that direct usage billing is rejected, it is clearly desirable to find some formula that will achieve at least a reasonably fair relationship between the use an institution is likely to make of a dataset, and the price it would have to pay if it chose to subscribe.

Past usage

The JISC has data for institutional usage of datasets in the past. But while this is probably a good predictor of future usage of the existing datasets (which tend to be research-oriented), there are plans for a much wider range of interests; so the usefulness of the existing data is limited.

Staff/student numbers

Because the JISC sector ranges from the smallest FE colleges to the largest research institutions, any formula for charging based on student and staff numbers ought to take account of the type or level of person being counted, and their potential interest in the dataset concerned. Given the current, largely research-oriented, datasets, for which we have usage data, it is possible to calculate weightings for student and staff types, and thus build a head-count model which does closely match actual usage.

But a model capable of being generalised to all possible datasets, should consider the probable pattern of usage of any given "dataset type" by any given group of staff/students. Classifying datasets in that way has proved almost impossible. Consequently, the model we have been able to build has had to rely on the assumptions that all datasets have the same usage pattern, and that all types of students use them equally. This does create some anomalies, illustrated by the following sample bandings and charges. (These examples assume a dataset for which the JISC wishes to recover £100k p.a., to which 50 institutions, spread across the sector, are expected to subscribe)

Central funding

Figures for the grant funding of institutions, by the HE and FE funding bodies and by the Research Councils, are readily available and reliable, and building a model based on such data is therefore simple and transparent. It also, unsurprisingly, produces a ranking of institutions which closely matches an ordinary, intuitive perception of the usage those institutions might be likely to make. Examples of bandings and charges for this method are shown below (using, again, the same £100k / 50 institutions assumption).

Band	Sample Institution	Band Price for dataset costing JISC £100k p.a.
A	The University of Cambridge	5,500
B	The University of Bristol	3,900
C	Nottingham Trent University	2,400
D	The University of Central England in Birmingham	1,750
E	The University of Kent at Canterbury	1,100
F	Dundee College	660
G	Pembrokeshire College	430
H	Harper Adams University College	300
I	London Business School	180
J	Orkney College	70

It should be noted that, while such a model charges directly on the basis of central funding, it is not at all the same thing as further "top-slicing", because institutions can choose whether to subscribe to any given service; it is simply that, when they do decide to do so, they would be charged on the basis of their size, as measured by central funding.

Total income

Given that some institutions, particularly in the "old" HE part of the sector, have considerable external income from other individual sources, and that student-based fee income varies widely, it might be thought that a model based on total income, rather than on central funding, would be fairer. However, while such total income figures are readily available, from HESA, for the HE part of the sector, it seems unlikely that such data could be readily obtained for FE.

However, for the HE part of the sector only, an experimental model was built, based on this total income data, and compared, for those institutions, with the one based on central funding. Perhaps surprisingly, the rankings in the two models were found to be extremely similar. It therefore appears that, even if the total income data were available for the whole JISC sector, it would add very little to what we already have in the central funding-based model.

3. Conclusions

Of the possible contenders for a charging model (i.e. those based on either staff/student numbers or on central funding) the model based on central funding appears fairer. This funding was endorsed by participants at the JISC's consultation event 18.5.01, and will be implemented for dataset agreements from Autumn 2001 onwards.