Let's take a hypothetical example for perspective. Suppose the Treasury decided to fund half the federal debt with floaters. Currently, the average interest rate on the federal debt is 2.8%. This would lead to some immediate savings. However, what happens if rates jump -- which is not unreasonable given the Federal Reserve's balance sheet of $2.8 trillion, or 19% of GDP. The savings would be wiped out if rates rose to 2.8% (which by the way, is roughly today's inflation rate). If rates go to 4%, then the government has to come up with an extra $1 trillion to pay the extra interest they precommitted to. Do you think 4% rates are out of the question? I certainly don't.
The uncertainty about future interest service is what I call funding risk. The amount of cash needed is unpredictable. If interest rates rise, then extra interest service must come from: 1) higher taxes; 2) spending reductions; 3) more borrowing; or 4) printing money. Note that (4) usually causes rates to increase even more, leading to even higher service costs. (3) can have a similar effect.
Yes, FRNs extend maturity -- if they are replacing securities that have a shorter maturity. Suppose they replace Treasury bills. In this case, the interest rate risk is identical (both pay the Treasury bill rate). Suppose they replace fixed rate coupons. In this case, the interest rate risk changes. As rates go up, the FRNs become more expensive to service.
The Treasury's February 2012 document is a good read. They make the case that there could be some cost savings in minimizing the rollover risk. That is, instead of going to market every three months for Treasury bills, you commit for a longer period, say, two or five years. They also argue there is a modest term premium that could be captured.
If the FRNs are just two years to maturity, then not much additional risk is created. This is especially true today when the Federal Reserve is on record saying that rates will remain low through 2014. However, what makes me nervous is the phrase: "... eventually the Committee anticipates FRNs of longer final maturities."