Premiums And Discounts: Flawed Thinking

October 23, 2009

Matt’s ideas for fixing bond ETFs aren’t bad, but focusing on premiums and discounts is a mug’s game.

Matt, I know you love it when issuers come up with creative ways to game the arbitrage process to increase liquidity and narrow tracking error, but the problem here is really more fundamental. The way I see it, bond ETFs will always be fighting two uphill battles.

The first problem is just woolly thinking with regard to premiums and discounts. The reality is that at any given point in time, the true NAV of any portfolio (ETFs included) is a moving target. The way I think about it, there are actually three claimants to the title of “true NAV”:

  1. There’s the rollup of all the “bids” in the underlying securities. This would be what the ETF basket would sell for if you could shove the portfolio through the top of the market.
  2. There’s the rollup of all the “asks” in the underlying securities. This is what an AP would pay if he had to go buy the basket to deliver to the issuer to make new ETF shares.
  3. There’s the rollup of the last traded price on the tape for the underlying securities. This is what gets used to create the NAV at the end of the day.

Each of these NAVs is actually completely fictitious, of course, but they’re the best we have. They’re fictitious because they don’t represent the true market value of one share of an ETF, or the true market value of the entire ETF. The fact that the NAV is essentially always wrong is what allows market makers to stay in business.

Think of it this way: If I have 1,000 shares of IBM stock, it’s not worth exactly the same amount per share as a 100,000-share trade printed last night at 4 p.m. It’s worth whatever I can actually sell it for when I go to sell it. Yet I measure my portfolio against a collection of these best guesses. When we talk about premiums and discounts in any ETF, we’re usually looking at these end-of-day NAVs compared against the end-of-day bid/ask midpoint for the ETF.

Honestly, unless there is a wide disparity between these two (which we’ve certainly seen in some ETFs, notably the U.S. Natural Gas Fund (NYSEArca: UNG) and some high-yield funds like the iShares iBoxx High Yield Corporate Bond Fund (NYSEArca: HYG)) or there’s a consistent premium/discount, I think it’s a bit silly to get hung up on the numbers.

Trends? Sure. But a day or two here or there where there’s a 1 or 2 percent gap between two numbers you can’t trade at? Sorry, not going to sweat it.

The second problem is simply that these funds are bond ETFs. Bond funds have to deal with the same NAV issues as equity funds, but with far more underlying uncertainty. Not only are they extremely optimized?as we’ve mentioned previously in our discussion this week?but the prices for bonds themselves are largely a matter of voodoo and tea leaves.

For something like Treasuries, it’s no big deal?the bonds trade constantly. For corporates, issuers are almost entirely reliant on pricing services that make up imaginary prices for bonds that haven’t traded in minutes, hours, days or weeks based on models that combine the current yield curve, credit ratings, recent trades, sunspot activity and the price of tea.

Again, I’m not saying there’s a magic solution. I actually think the current system works remarkably well. But bonds are simply noisy.

Add the noise of bonds to the noise of doing premium/discount calculations, and the idea that you can get any kind of real, short-term information from a spot-check seems spurious. Bond investors should be thinking long term, and their analysis of bond ETF spreads and discounts should take the long view too.

Find your next ETF

Reset All