The maths is fine of course. The same analysis was used by many people to decide it would be a good idea to borrow money to build lots of houses - in places like the US, Ireland, Spain and so forth. The outcome, as we all know now, wasn't terribly rational. The maths isn't limited to self-improvements - any kind of improvement or construction activity has the same economics.

There isn't anything wrong with the mathematics. The difficulty is that it requires us to speculate on what the future looks like. Low interest rates dramatically increase the importance of the further-away future - if you halve the interest rate, you (more or less) double the time into the future that is relevant to your calculation of whether your activity will be profitable. And the future is uncertain, as all post-crunch property developers (and their bankers) know. The prevailing low interest rates encouraged the speculation into the far future that we now know is different from that envisaged.

Corporations are (contrary to AlephNeil) potentially immortal(ish), motivated by the net present value of their future earnings, and don't like switching careers. They are notoriously short-term planners in many cases, and uncertainty is the reason why. Most think that improvements should pay off in 3-5 years, or they aren't worth doing. It's not the prevailing interest rate that counts here.

Self-improvement is the same - it's pretty uncertain - you don't know what you're going to get for it. Well, in the less interesting cases like gaining a qualification, you might have a reasonable estimate. But when doing self-improvement at the forefront of human knowledge, you seriously don't know what you're going to get. But that's by far the most interesting place to self-improve, I'd say.

I guess everyone on this blog is a self-improver almost by definition - those who don't improve just because they like to won't find this place interesting, and probably won't understand it either. I think most of us here will therefore already have answered the question, and had some experience of what that answer leads to.