I’m not aware of any costings which have actually been done, but my gut feeling is that the balance digital vs. paper comes out in favour of digital. (By “paper” I’m including parchment, photographs etc too.)
1. The biggest single ongoing cost in any repository is staffing. A paper-based archives service has to run searchrooms for users to consult the materials, where users are supervised and security is ensured. So, paper-based repositories have to employ receptionists, searchroom assistants, relief staff to cover when other staff are away etc. A digital repository which makes its assets available over the web does not incur any of these costs.
2. The next biggest direct cost is the building itself. Repositories which meet BS5454 or other relevant standards for paper preservation are not cheap things to build or run: they require space for literally miles of shelving, environmental conditions within fine tolerances, constant monitoring, security systems, controlled searchrooms and so on. Then in addition, they need machine rooms for their large ICT kit as well, like their servers. But a digital repository is just one big machine room, really.
3. There is a disparity between future growth costs. Electonic storage becomes more efficient all the time: costs of storage come down. The costs of paper storage, on the other hand, go up as more paper documents are added. (Where I work we are planning to build two entirely new repositories just to cope with all the paper which we receive, at a cost of millions! Whereas even enormous servers only cost tens of thousands.)
4. There is a disparity between location costs. Because paper repositories depend on users visiting them to consult the documents, they tend to be built where transport links are good, which in turns means that they are usually built in cities or major towns. Paper-based repositories therefore exist in areas of high land values, high business rates etc. A digital repository which makes its assets available over the web can minimise these costs by siting itself in an industrial unit pretty much anywhere.
5. There is perhaps an indirect cost to society as a whole. Paper repositories expect searchers to visit them. These searchers have to take the train, drive or fly to the repository, creating a transport and environmental cost. They have to visit the repository when it is open, which means taking time off work, creating a minor hit on the overall economy. These societal costs are minimised by digital repositories.
So, where does the perception come from that digital preservation is actually very expensive? Perhaps:
digital preservation may involve complicated software technical work, ie. it involves techies, and employing techies is seen as expensive (whereas employing professional archivists isn’t)
but more probably it’s perhaps
the repositories undertaking digital preservation are already running paper preservation. So, digital preservation is an additional cost on top of the paper preservation costs already in place, and therefore it is seen as expensive. A paper repository may cost (say) £500,000 a year to run, but because it has always cost that, the finance people accept it, and can budget for it. The digital preservation function may only be £100,000 a year to run but because it is new, and on top of the paper cost, the finance people get sweaty and worried.
Anyway these are just my thoughts! Let’s see what happens in reality when the cost comparisons are actually done.