That's the thing though, the 10% number is fuzzy. I agree there's definitely a productivity boost from having a private office, but it's hard to quantify. The smaller square footage and lower lease payments are easy to quantify. If productivity expectations are set based an open office, then the increase with the more expensive private offices is unknown.
Additionally, high overhead increases risk. Maximum productivity is only important as long as you have more work than there are employees to do it. What if there's a downturn (as in 2008)? Now you're locked into an expensive lease and no longer have the revenue to support it. The company with lower overhead can hold out longer until things turn around. Employees can be laid off. Getting out of a lease or selling a building are harder.
The thing is it doesn't have to quadruple in order to justify the extra real-estate cost because your monthly salary is (probably) much higher than the monthly rent on your desk space.
If you're paid $120k (10k/month) and rent is $5/sq-ft per month [1]; going from 36 sq-ft of space for your desk (small) to 144 sq-ft of space (4x as much) will cost $540/month. Meanwhile even a 10% productivity decrease would cost ~$1k per month.
Edit: Thinking more about it, the productivity loss is even worse than above. You might be paid $120k/year, but the company expects to extract some multiple of this number in value from the work you do (I've heard target numbers like $500k/year) - so a 10% decrease in productivity is more like $4k/month in lost value.
[1] https://www.quora.com/How-much-is-typical-office-space-per-s...