Not only 'in theory', but it is used for ancestry testing.
Coding DNA (exonic regions) tend to be more highly conserved between individuals, as changes to the sequence are more likely to have a deleterious effect on an important phenotype (think a mutation which makes someone only have a single arm). Non-coding DNA is less constrained by selection and varies more between individuals / ethnic groups. This makes it more useful for identifying variants specific to different ethnicities, which makes the basis of ancestry testing. Typically in ancestry testing SNPs which show evidence of being under some kind of functional selection (i.e. are out of Hardy-Weinberg equilibrium) are excluded from the analysis.
I'll try and update this post with some references.