The "cost of the product" is around 10 bucks for the box, disc, manual and licensing fees. The big problem is R&D but that's fixed cost, you cannot give a price that will gurantee the game will break even (as opposed to selling, say, forks where your cost is material and labour and only changes for large quantities), you can only say "at this price we need to sell this many units to break even", compile the price->demand curve and look where you might make the most money. Many forget the price->demand curve but some care so they price games they don't expect to sell well lower (e.g. Serious Sam, Viewtiful Joe, Alien Hominid). Some managers are completely delusional about the quality and therefore demand for their product.
This problem leads to the hit-driven business model we're seeing today, games carefully aimed at certain demographics with hopefully predictable sales and therefore calculable price->demand curves so they can predict how much money they can spend on the development to get returns. Of course too many ignore the primary factor here, quality. You can have the best idea but as the Heavenly Sword diary says, "you cannot copy good execution". Innovation adds another uncalculable factor and managers love to pretend a problem is predictable (even though quality and innovation are both pretty random and based on the skill of your devs, no innovation doesn't mean they'll make a good game).
(note: Usually it's called "supply/demand curve" but few items are really priced by supply at retail except if they overstock and don't sell. Competition works differently in the IP market as well so competitors dropping the price will not necessarily lead to less sales for you.)