If a university wants the work to count towards the targets it is measured by by government, or if it receives certain types of public funding (which is pretty

If a university wants the work to count towards the targets it is measured by by government, or if it receives certain types of public funding (which is pretty much most of it), there's an expectation in the UK for it to be open access within agreed timescales. If it's not, then it can't be counted for the key reporting metrics. The end result is that the vast majority (i.e. almost all) such research (regardless of how funded) is public and accessible on the institution's repository, or otherwise deposited with an open access library, such that it can "count" towards the metrics. A clever approach if you ask me - find what matters to the individuals concerned (the metrics), and discount work from those metrics unless made freely available within X days, with appropriate audit trail to match.