20 years is far too short.
The introduction of UBI divides society into the "pleasantry", who live modestly on the UBI, and the "workers" who aspire to a higher standard of living, study to get qualifications, and then work. Take home pay has to be high to get people to work at all, taxes have to be high to fund the UBI, therefore gross pay has to be very high. Let us assume major advances in automation, so that the sums work out.
In the next generation there is a class divide. Children of workers get brought up with a set of expectations about studying in school, getting qualifications, and working twenty hours a week at a very highly paid job (perhaps repairing broken down automated systems - endless one-off jobs that are ill suited to modern statistical artificial intelligence). Children of the pleasantry grow up with a different set of expectations around being thrifty and enjoying simple pleasures (simple is perhaps a euphemism for cheap, but there is also an aspect of living a gently paced life that passes copious spare time enjoyably).
I anticipate a two-way flow. Some of the children of the workers struggle academically and see no prospect of ever learning to trouble shoot a broken down robot. They give up and grow up to be members of the pleasantry. Other children of the workers just think that the pleasantry have the better deal and drop out of the working class. Meanwhile there is traffic the other way, as children of the pleasantry query why other children have expensive toys. The adult world is explained to them: the children with expensive toys are the children of workers. And so some children of the pleasantry learn to aspire to more, to study, work, and have luxuries.
The key issue is the numbers. If 50% of the children of the workers join the pleasantry and 10% of the children of the pleasantry join the workers, then the working class will shrink and the pleasantry grow.
For example, maybe the first generation sees an even split, 50% workers, 50% pleasantry. The proportion of workers falls to 30%, then 22% then 19%,..., solving for the equilibrium, we only have 17% workers, even though the 20 year trial had the proportion of workers holding up reasonably well, above 30%.