The Norwegian expert commission (1) used a meta analysis by Grotenhermen et al (2) to determine the blood level of THC that is equivalent to a BAC of 0,04 %.
As promised in The Edible Mistake (1), there is more to say about route of administration and the legal limits for impairment. I will start out by providing a quote from Grotenhermen et al regarding the authors assessment of the findings that were used to determine the equivalent to BAC 0,04 %.
"Note that the correlation between THC serum concentrations and impairment did not depend on the route of administration of cannabis (inhalation, oral ingestion)."
We can infer from the graphs below that the equivalent is artificially low. This occurs in two dimensions. When cannabis is ingested, low levels of THC provides a significant and sustained impairment. When cannabis is smoked, high levels of THC provides impairment, while low levels of THC occurs when the user is unimpaired. Depending on how the studies are made, many factors come into play when this skews the relation between blood levels and impairment.
The selection where cannabis is ingested inflates impairment at low blood levels relative to smoking.
The selection where cannabis is smoked deflates impairment at low blood levels relative to ingestion.
In culpability studies there might be a time gap between a traffic accident and blood sampling. This leads to attribution of accident risk to artificially low blood levels. Especially when smoking.
Long storage times of blood samples might lead to degradation of THC and attribution of accident risk to artificially low levels.
Culpability studies can be skewed by residual levels of THC that is interpreted as a cause for accident when the true cause is human error unrelated to impairment.
Blood levels edibles 
Impairment edibles 
Blood levels smoking 
Impairment smoking