One of the logical distinctions that needs to be made in thinking about the purely hypothetical effect on crime of a prenatal genocide of an entire ethnic group is:
Are we talking about what would be the impact on the total number of crimes in the country?
Or are we talking about the impact on the national per capita crime rate?
Steven D. Levitt, author of the abortion-cut-crime theory, tries to glide past the nasty racial implications of his theory by claiming on his blog:
... if you prohibit any group from reproducing, then the crime rate will go down)...
But that's not true. If all future Asian-Americans were aborted, the national crime rate, as measured in per capita terms, would go up because the Asian-American crime rate is below the national average. (Asian-Americans in 2001 were incarcerated per capita only 22% as often as whites and only 3% as often per capita as African-Americans.)
For ethnic groups with higher than average crime rates, the opposite would be true.
Now, please don't claim I'm advocating genocide. Indeed, for six years, I've been a voice crying in the wilderness saying that Levitt's theory that abortion-cut-crime turns out not to be true when you look at the actual historical record in any detail, which Levitt failed to do when he concocted it.
For an explanation of one reason, besides the crack wars, why the black violent crime rate shot up among the cohort born after legalization of abortion, see here.