Riskless curve

I read this paper by Longstaff, Mithal and Neis (2004) in which they find the choice of the riskless curve has a big impact on determining the components of credit spreads. Academics typically use US Government Treasury securities as the benchmark riskless rate, however, Longstaff et al. (2004) use the swap curve instead. They find, in contrast to many previous studies, that default risk in fact explains most of the variation in spreads.

Now this is interesting as it casts doubt on previous results. It could potentially be that previous studies, using swaps rate instead of Treasury rates, might find that default risk is the major determinant instead.

The question is, why is the swap curve a better proxy for the riskless curve? Hull et al. (2004) provide an argument, but I am not thoroughly convinced (at least not intuitively clear) that is the case. They also mention that practitioners also use the swap curve as the benchmark.