Based on basjoos' comments, and seeing several concept cars with front wheel skirts, I tuft tested and then even went ahead and built one front articulated skirt last fall. See:
Video: tuft testing front wheel skirts (Geo Metro)
Designing/building front wheel skirts v.1.0 (Geo Metro)
The tuft testing video certainly seems to show the potential benefit of skirts.
However, today I went out and did some coastdown runs, A-BBB-AA-BB with a pair of easily attached/removed cardboard skirts, and the results seemed to suggest that the style of skirts used actually
increases drag slightly. EDIT: but the difference compared to the standard deviation of the data sets is so close as to be statistically meaningless.
Bummer, but based on this, into the waste bin they go!
I'll post the full data tomorrow.