The EIA Annual Energy Review 2008
http://www.eia.doe.gov/emeu/aer/ estimates that 1340 trillion Btu was used in the US in commercial buildings for lighting during 2003 (p.65), out of a total of 99.30 quadrillion Btu total US consumption in 2008 (p.3), of 18.54 quadrillion Btu total commercial use (of which 1.34 quadrillion Btu is 7.2%, which is, as you say, less than 20%). US residential usage amounted to 21.64 quadrillion Btu in 2008 (p.3).
Suppose you're right and 11% of typical residential usage is lighting (what's your source for that, please? The AER doesn't seem to hazard a guess there), and that the US isn't too atypical (admittedly, a very dubious assumption). Then that would amount to 2.4 quadrillion BTU spent on residential lighting in the US.
The total, then, would be 1.3 + 2.4 = 3.7 quadrillion Btu per year used on residential and commercial lighting, or 3.7% of the total.
Industrial facility lighting adds another 0.2 quadrillion Btu (p.48), which would bring it up to 3.9%.
It would immediately follow that 3.7% of the average person's energy use is residential and commercial lighting, and 3.9% is residential, commercial, and industrial lighting, which is indeed a relatively small proportion. If every person in the US went entirely without artificial lighting but otherwise somehow continued their lives as before (except in cars, which are not included in the above statistics), total US marketed energy consumption would drop by 3.9%.
I am in favor of compact fluorescent lights, ordinary fluorescent lights, halogen lights, and new high-efficiency LEDs. But the savings are not enormous when considered overall as a society, particularly considering that most industrial commercial lighting is already using more efficient machinery than incandescent bulbs. They are, rather, relatively small.
3.9% of US marketed energy consumption is 130 gigawatts. 3.9% of world marketed energy consumption (about 500 quadrillion Btu/year, according to IEO2010 Highlights, p.1) would be about 700 gigawatts.
Laws are not admissible as evidence about energy usage. Laws provide evidence about political reality, not objective reality.
Do you not "share the view" that 3.9% is a relatively small proportion of 100%? I think that would merely mark you as a Humpty-Dumpty. (There's glory for you!) Or do you think that the US is deeply atypical, and that the average numbers are much higher than the US numbers?