In my country, there's a system to reimburse companies for their costs after a certain number of sick days–3 weeks I believe. That protects smaller companies, who might otherwise get into serious trouble if, say, one of three employees can no longer work but still gets a salary.
Of course there are trade-offs involved. It does cost some money, and it does add a bit of bureaucracy. Some abuse may also happen, although by and large it's rather low–there are some 'social' control mechanisms, and the data pretty clearly shows that absenteeism is directly linked to employee satisfaction.
My suggestion, while obviously extreme (flexible scheduling as mentioned elsewhere in this thread would be a much more reasonable 'solution') wouldn't fly in the USA as nobody takes their holiday days anyway!
A friend of mine is a (non-computer) engineering executive and has a particularly generous PTO package of 200 hours a year with a 400 hour rollover cap. He keeps a bank of about 250 hours in case he needs them, but other than that he's using it as fast as he gets it. This in addition to the 12-14 holidays he already gets.
But implying that the average yearly leave is 0-5 days including sick time is just flat out wrong.