lack_ey wrote:I think I'd rather just hold Treasuries myself if that's the exposure I wanted, but okay. There's definitely a market for this stuff.
That's interesting. You would do that even if you wanted to maintain a similar duration bond portfolio, for a small portion of your portfolio (I imagine you would use some kind of ladder)? I hold about 10% in VGIT. I'm not really familiar with what the spreads are on treasuries for implementing something like this. If it is truly cheaper I would consider it.
First of all, I have to say that I don't have enough money in Treasury bonds for this to be worth my time, so I don't do it personally.
You can get Treasury bonds at auction, so if you just did a ladder and bought new issues on maturity, there wouldn't be spreads to deal with. Alternatively you could just target some range and hold a few issues, like buy a 7-year Treasury every year and sell anything that gets down to 3 years (so you might hold 7/6/5/4 and then in a year that's 6/5/4/3 whereupon you sell the 3 and get a new 7).
That said, some off-the-run issues may have slightly better yields, better than the average of what a fund might hold. There are typically small differences in yields between Treasuries based on investor preference for liquidity and then some small gap that can't quite be closed by the pair traders.
Typical spreads in the secondary market are around 0.03 (quoted referenced to par of 100). So around 3 bp. Spread out a couple of transactions over a few years of holding and this is pretty small. Keep in mind that VGIT has a spread around 7 bp, though you might not be trading that as much as individual bonds in a ladder you could be managing.
Downside is obviously your time/effort (though this is low) and then the kinda-chunky transaction sizes, generally in the range of $1,000 per bond, sometimes requiring larger lots than whole numbers. Then having to think about what to do with the interest payments and investing those. Of course you can just stick the remainder in other funds or keep as cash, so not a big deal.