There is clear evidence that driving performance is impaired even at those levels.
Examining the statistics of traffic accidents supports the conclusion that 0.08% is clearly related to accidents:
BACs of .08 g/100 mL or higher were found in 38% of killed and 30% of injured drivers, 37% and 27% of passengers, and 35% and 22% of male motorcycle riders.
—source
More evidence, even for lower limits, comes from double blind experiments:
Statistically significant and meaningful decrements in driving-related performance were identified at 0.05% BAC or below in many studies. Younger and inexperienced drinkers and drivers appear to be a greatest risk for alcohol-related traffic crashes. It is concluded that on scientific grounds, there is support for setting the legal limit at 0.05%.
—source
This paper also confirms the findings.
Now, are BAC lowering policies effective? Not very, unfortunately. This has been studied in Slovenia:
Further concern arises from the finding that heavily drinking drivers are not responsive to penalty increases. We came to the conclusion that, in addition to raising fines and expanding penalty points, the introduction and effective enforcement of complex legislative measures together with wide community action are necessary to resolve DUI problems in Slovenia.
—source