Exchange-traded funds continue to be one of the fastest-growing segments in the investment world. According to the Investment Company Institute, investors pumped some $53.3 billion into ETFs for the year-to-date through August 2009.

While they are a great tool, advisors must proceed cautiously when considering ETF strategies, such as bond ETFs and more exotic offerings, such as commodity tracking funds, leveraged ETFs and Exchange Traded Notes.

If we start with bond ETFs, we can see that a number of these funds have traded at significant premiums (or discounts) to net asset value over the last year.

Premiums and discounts to NAV are unwelcome, insofar as they cause the returns that an investor earns to vary from what the fund's underlying investments earn.

For instance, a prominent high-yield bond ETF was trading at a very steep 11% premium NAV at the start of the year. But as of Oct. 6, 2009, the ETF was trading at a far smaller 0.3% premium to NAV. When a fund trades at a shrinking premium to NAV (or an expanding discount to NAV), its market returns will lag NAV returns. Consequently, the ETF's year-to-date market returns (19.9%, as of Oct. 6) fell well shy of its NAV returns (32.5%).

Though some observers argue that this is a logical outcome-stemming from differences in the liquidity of bond ETFs and the underlying fixed-income securities in which they invest-we at Morningstar are not entirely convinced. Based on our research, liquidity alone doesn't explain the way some bond ETFs have traded.

Consider, for instance, a well-known ETF that tracks a national index of municipal bonds. While this ETF's price has frequently strayed from its NAV, these divergences haven't been uniform. In fact, when we examine the rolling 30-day correlations of the ETF's price to its net asset value, we find that the gap between them has widened and narrowed like an accordion. While it is true that the municipal bond market can become more, or less, liquid from time to time, we wouldn't expect liquidity to whipsaw to such a degree.

This leads us to conclude that other forces are at work. Such uncertainty is obviously not welcome. We want to be able to invest in an ETF with utmost confidence that its price will correspond closely to NAV.

We've also been nonplussed by some of the more exotic ETFs, including commodity-tracking funds, leveraged ETFs, and exchange-traded notes (ETNs). Take, for instance, a very popular ETF that invests its assets in the near-month, crude-oil futures contract in hopes of approximating changes in the spot price of crude oil.

The problem? The fund's returns have scarcely resembled the change in spot prices due to the shape of the futures curve. (To get more precise and technical for just a moment, when contracts in future months become more expensive than the near-month contract, an upward-sloping futures curve, known as "contango," imposes a cost on investors when they roll the expiring near-month contract over to the next month).

While the shape of the futures curve is technically not the fund's fault, as contango and "backwardation" (i.e., a downward- sloping futures curve) are recurring phenomena, there's some reason to suspect that the fund has become a victim of its own success. That is, it's possible that investors, who recognize the ETF's huge footprint and its need to roll contracts on a monthly basis, are bidding up contracts slated to expire in future months (i.e., the contracts into which the ETF will eventually roll).

Sadly, this has not been an isolated occurrence. Other commodity ETFs, including a prominent fund that invests in natural gas futures, have been roiled by constraints that regulators recently imposed in order to limit the size of participants in the commodity futures market. In addition, the issuer of a commodity-linked ETN recently liquidated the note for similar reasons.

The common thread in these examples is a mismatch between an ETF's structure-which demands liquidity to facilitate intra-day trading and ensure that the fund's market price approximates its NAV-and the size and underlying liquidity of the markets in which they're investing. When the underlying securities are insufficiently liquid or the market isn't large enough to support an ETF, we tend to see dislocations of the sort described.

Leveraged ETFs are in a class of their own. These funds are tied to the daily returns of a reference index. Consequently, they are suited to short-term traders, not long-term investors. We will not invest in leveraged ETFs, long or short, which are tied to daily returns. Therefore, as a practical matter, the scarcity of unlevered inverse ETFs makes it more difficult for us to construct strategies that opportunistically short or hedge various market segments. (It's worth noting that there are other ways to gain inverse market exposure using ETFs. For instance, shorting an ETF or buying put options on an ETF.)