If you pre-approve but not fund certain desirable projects, and allow unused funding or surpluses gained through efficiency to spent on those, you get the double-win of maintaining trust and making efficiency organisationally desirable.

I agree that trust is part of it, but I don't think it's the whole story. A lot of this is also due to cognitive biases. For example, optimism biases make public works project managers continually underestimate cost and schedule. Then sunk-cost biases make funding managers scramble to find additional money to make up the difference. Often, those funds come in the form of "unspent" funds from other projects.

That's the generous interpretation. There's also the case when project managers deliberately low-ball their project estimates to take advantage of the phenomenon. I'd put that in the trust problem category.

It seems more like an incentive problem. I'd be interested to see what happens if you give the deciders X% of any unused budget as an EOY bonus.

While SharePoint is not the best choice for many applications, it has enough features to make it work. Since there are plenty of SharePoint experts already in the organization, it could be cheaper to just leverage them to build some SharePoint site that just barely works well enough for whatever application. It would cost a lot more to bring in some consultant that knows how to setup the proper software to do whatever new thing the project needs to do along with whatever new service contracts that need to be signed and paid for to cover the software for years. Do you jump into the new software that might be better or do you just use the same old SharePoint that you've already invested heavily in? The efficiencies of both solutions are probably intangible and it would be difficult to price either of them. Eventually someone out there will be the evangelist to convince program management that the new software really is the best choice (Jira, Confluence, Teams, Salesforce, Trello, etc).