I suspect that part of that may be market growth beyond expectations. If telcos were bidding about the same net present value for the one time auction and the rev-share auction, with their revenue projections and revenue came in higher, that's going to be better for the government.

It's also easier to bid higher in rev-share, because you don't need to finance that net present value; it's just going to be built into the price of service, and your competitors will likely have similar rev-shares (depending on the auction process), so it's going to be built into everyone's prices.

I think there's also some value in a stream of payments rather than a one time payment. Although you can exchange one for the other with financial tools, a government may find a stream of payments to it matches well with the stream of payments it's making.

If the government auctions off spectrum at a flat price then it gets certainty, which normally comes at a premium; doing revenue share the government retains more of the risk, so orthodox economics would say that it will get a better price on average.

There's also an argument that the government doesn't actually realise any advantage from selling off the risk: if the telecom companies fail, it's still the government's problem. So the government ends up paying extra for something it can't actually use.