We Love This ETF: Sell It Now

Something strange appears to be going on with municipal-bond exchange-traded funds. The two largest ETFs in the category have had very different performances for the year to date.  IShares S&P National AMT-Free Municipal Bond (MUB) has returned 4.90% this year. Its main competitor,  SPDR Nuveen Barclays Capital Municipal Bond (TFI), has returned only 2.19%. Both ETFs track investment-grade municipal bonds, have similar yields and durations, and have fees that are within 2 basis points. From a holdings perspective, the only material difference is the slightly higher credit quality of TFI, but even that difference is negligible.

Why would similar funds have such different results and what should investors do? Upon further investigation, we find that what appears to be strange is in fact what we should expect, and the course of action is simple.

It's Not You, It's Me
I am an advocate for well-constructed funds, so it pains me to say this: It's time to part ways with MUB. This recommendation has nothing to do with our outlook for the muni-bond market. It just boils down to returns. MUB has been trading at a substantial premium to its net asset value lately, a phenomenon more frequent in closed-end funds than in the ETF marketplace. However, TFI is still consistently trading near its NAV. So, if the fund characteristics are so similar yet one is trading at fair value and the other is trading at a 3% to 4% premium, why wouldn't you swap them? The premium is almost double the current SEC yield of approximately 2%.

I do feel slightly guilty about making this recommendation since the iShares fund has served investors well. In fact, it has served as a better tool for accessing the muni-bond market than TFI. The ETFs have had very similar performance until recently when MUB's returns surged ahead. Once the premium on MUB disappears, which will most likely happen in the near future, feel free to swap back if you'd like. You'll set yourself up to capture the premium (or discount, depending on market conditions) again in future years.

You have to venture into the ETF sausage factory to see why the performance gap has emerged recently, and the research is worth the result. You'll find that it is completely logical for temporary but sizable premiums to occur on MUB while TFI stays close to its NAV. However, you'll also find that MUB will likely have better NAV performance over longer periods of time because the fund incurs lower trading costs. These lower trading costs could manifest into substantially higher returns if TFI realizes substantial fund inflows during periods when the muni-bond market is trading inefficiently, as it is today.

Municipal Market
All ETFs trade as a function of their underlying market, and the current municipal-bond landscape is supply constrained. In January there was $7 billion in inflows to muni mutual funds and ETFs, which is a turnaround from 2011 when there was $9.8 billion in redemptions. To make matters worse, new issuance of individual muni bonds in January was anemic. Thus, we've seen outsized demand by new investors precisely when supply of new bonds has been limited.

Another factor that is limiting liquidity is that some major muni-bond dealers are holding less inventory than they have in the past. The infamous "Volcker Rule" seeks to reign in risk-taking by limiting the proprietary trading at banks. Maintaining an inventory of bonds available for sale is considered proprietary trading so the major dealers have started reducing the size of their inventory in anticipation of the new rule. While many speculate that the rule will not be implemented in its current form, the fear of the law's current structure is enough to curb market participation. The net effect of these three factors has created a situation where the bid-ask trading spreads on bonds have widened dramatically.

The Sausage-Making Process
The process that keeps an ETF trading near its net asset value is the creation/redemption process. Without the creation/redemption process, ETFs would simply be closed-end funds. Institutional trading firms called authorized participants are the ones who work with the ETF provider to handle this process. If there is strong buying demand for an ETF, then it will trade at a premium to its NAV. However, when the premium gets to a critical level, it will become profitable for an authorized participant to arbitrage the premium away. To eliminate a large premium, more shares need to be created to satisfy market demand. This is when the APs aggregate and deliver a predetermined basket of bonds, called a creation basket, to the ETF provider. In return for the basket of securities, the ETF provider gives back shares of the ETF. The AP then sells the newly created ETF shares to buyers on the exchange for a slight profit. With the extra demand for shares satisfied, the ETF will trade back near its net asset value.

MUB and TFI both hold hundreds of different bonds in their portfolio. It would be very difficult for an AP to buy up hundreds of illiquid bonds to do a simple creation, so ETF providers typically have a creation basket that is substantially smaller than the index. For an ETF with 500 securities, this creation basket may consist of only 10 to 40 bonds. Under normal market conditions, it is relatively easy to buy the bonds in the basket. But as we identified earlier, the current supply-constrained market has made it much more difficult.

In an illiquid market, fixed-income ETFs can trade at substantial premiums because of increased trading spreads. ETFs calculate their net asset value based on the market "bid" price. To create new shares to meet demand, APs have to buy bonds at the "ask" price of the market. This difference between the bid and the ask in the muni market is well over 2% right now. If APs are paying 2% more to create shares they are going to pass that difference down to investors and you'll see it as premium in the ETF.

Two Different Approaches
Relative to the rest of the industry, the folks over at iShares are the staunchest ETF purists. Whenever possible, they stick to a consistent creation/redemption process where they receive only physical underlying shares for their funds--along with enough cash to make up for rounding differences and accumulated dividends or interest. However, not every firm follows this model. Many ETF providers operating in less-liquid markets will accept cash in lieu of securities, thus performing a "cash creation."

Cash creations are nothing new. (You could argue that every open-end mutual fund "creates" new units through the cash process.) When State Street teamed with Nuveen to create TFI, they had an eye on flexibility. From time to time, the fund provider will accept either securities or cash from authorized participants to create new shares. Why do they allow this?

Because it is so difficult to aggregate muni bonds right now, authorized participants have to pay considerably higher prices than the index actually indicates they are worth. Thus, the flexibility lowers costs for the APs. This results in the ETF trading at a smaller premium to NAV than a fund that uses in-kind transactions.

On its face, it seems that allowing flexibility in the creation process has a lot of upside. In reality, the flexibility is a zero sum game. The perceived benefits are really just a transfer of wealth from one set of investors to another. In our case, investors who purchased TFI several months ago are subsidizing investors who choose to buy it today. This is one of the few cases where even a free-market proponent should love subsidies.

The cost of purchasing the muni bonds is simply shifted away from the APs and onto the owners of TFI. Any new cash that flows into the fund will eventually be used to purchase muni bonds on the open market. If we assume that TFI's traders are equally efficient at trading bonds as the APs are, then the net purchases will occur at a premium to the index level. Thus, each incremental dollar put to work will cause the fund performance to lag by the amount of the premium.

When TFI accepts a cash creation the portfolio manager will need to go the market and buy municipal bonds directly. If the market is illiquid the manager will have to pay the elevated ask price of the market. This premium is a cost borne by all shareholders in the fund instead of just the new purchaser in the case of the iShares ETF MUB.

Because muni bonds are trading inefficiently, they are exhibiting a hefty liquidity premium. If you are curious about how liquidity can impact returns, check out "Liquidity as an Investment Style" by Ibbotson, Chen, and Hu.

For How Long?
The final catch is an important one. In the highly likely event that the liquidity of the municipal-bond market reverts to its average, we would consider switching back to MUB. Why the flip-flop from the stodgy buy-and-hold folks at Morningstar? Well, it's all about performance.

IShares is very strict about the bonds it will allow in the creation basket which helps the fund maintain very accurate index tracking. If you look at Morningstar’s Estimated Holding Cost data point, MUB costs 0.33% per year to hold. Most of the holding cost is the expense ratio of 0.25%. If you exclude this cost, MUB has displayed a tracking error of only 8 basis points. This is tremendous tracking when you consider that MUB tracks a very broad index of more than 8,000 bonds. The downside to very accurate tracking is that premiums can develop in your ETF when the market becomes illiquid.

The upside to the in-kind creation process is that MUB will likely exhibit even better tracking than TFI over the long term. Because the high-cost of purchasing new bonds is not included in the index today but will manifest in NAV performance over time, we would expect this to result in a performance drag going forward if a substantial amount of funds flow in during this period of muni-bond market illiquidity. It's one of the few cases where we will be able to attribute predictable tracking error to fund flows.

	ETFInvestor Newsletter
	Want to hear more from our ETF strategists? Subscribe to Morningstar ETFInvestor to find out what they're buying—and selling—in their portfolios.	One-Year Digital Subscription 12 Issues | $189 Premium Members: $179 Easy Checkout

Disclosure: Morningstar, Inc., receives fees for licensing its indexes to ETF/ETN providers. These fees are mainly based on fund assets under management. BlackRock Asset Management; First Asset; First Trust; Invesco; Merrill Lynch; Northern Trust; and Scottrade currently license Morningstar Indexes. These ETFs and ETNs are not sponsored, issued, marketed, or sold by Morningstar. Morningstar does not make any representation regarding the advisability of investing in ETFs or ETNs that are based on the Morningstar Indexes.