All it takes is a cursory glance at almost any mass migration in the US to show why that amplifying effect doesn't exist: when the [x]% finally get the money to move in, everyone else usually moves because of prejudice, driving the value of the neighborhood down and the [x]% into bankruptcy or financial destitution. I can even name a few towns in California where this has happened in my lifetime.
That's the kind of feedback loop we're talking about. No one bases their prejudices on actuarial tables.
Jim Crow laws were enforced until 1965. They existed in the lifetimes of part of the current generation of would-be homeowners, and that is one thing you can't lay at the feet of the banks.
> All it takes is a cursory glance at almost any mass migration in the US to show why that amplifying effect doesn't exist: when the [x]% finally get the money to move in, everyone else usually moves because of prejudice, driving the value of the neighborhood down and the [x]% into bankruptcy or financial destitution. I can even name a few towns in California where this has happened in my lifetime.
The events you're describing should have the opposite effect. The first family to move in is obviously screwed, but screwed in a long-term sense where they lose the value they paid for their home which only happens in some decades when they want to sell it. Until then their mortgage payment is the same one they budgeted for to begin with. And all the racist white people who are leaving are in the same boat, but right away -- they want to leave but no one will buy their house for what it was worth last year.
Meanwhile there is suddenly a whole neighborhood where housing prices have plummeted and a bunch of white people who would otherwise come in and bid the prices back up are opting out. Doesn't that just make it easier for other black people to buy those homes for a discounted price?
The only reason that happens it that you're using exactly 100%.
If it's 98% then the top 2% in that group get approved, which gives you new data on that group that tells you your old numbers are inaccurate and a lower percentage of the group defected than predicted, so in the next generation you approve the top 10% in that group and so on.
This is the exact opposite of a feedback loop. As time passes it amplifies the signal, not the noise.