Now that Shuttle's development costs are far sunk, it is a much more affordable launch vehicle than Constellation's Ares, by almost any measure - total life cycle, average and marginal cost. The $2 billion average cost per flight of Ares 1 blew me away.
To start comparing Shuttle costs to commercial launch, NASA's 2008 budget for Shuttle was $3.2 billion and in 2009 was $2.9 billion. Divide that by an average of five flights per year and you're looking at $580 million to $650 million per flight.
However, if you disaggregate the requirements for delivery of crew and cargo given the demands at Station, the Shuttle lift capacity is oversize and the launch cost less affordable - once Station assembly is complete.
Compare with the $3.5 billion Commercial Resupply Services (CRS) for cargo delivery to Station from 2011 to 2016. NASA could try to keep Shuttle flying at a low rate instead of pursuing CRS, but, given the reference cost of the stand-down post Columbia (Shuttle was still burning up $2-3 billion a year without a single launch) the Shuttle would still not be cost effective given the operational resupply demand of Station.
For crew transportation to Station, using a conservative price of $50 million per seat on Soyuz, let's assume NASA needs to fly ten astronauts to Station each year. That would still only amount to $500 million per year at Russia's extortion pricing.
I suppose if NASA could minimize Shuttle program costs to below $1.5 billion a year from the remainder of Station operations, it could be the more affordable option.