Do ETF or Index Sector Funds Make Sense?

Updated April 2006
PDF April 2006 Journal

In markets around the world, sector and sub-sector exchange traded funds are rapidly growing in popularity. In the United States alone, their assets under management are up by almost $6 billion dollars (21%) so far this year. To capture these flows, many new products are being launched, based on even more narrowly defined indexes. Given these trends, this is an appropriate time to ask a basic question: does this make sense?

Let's start with a basic definition: just what is a sector index fund? In simple terms, these are based on an index that tracks the performance of a group of similar companies. Most of these products utilize one of two systems for classifying companies. The first is known as the "Industry Classification Benchmark" (www.icbenchmark.com). At the lowest level, the ICB identifies 104 groups of similar companies, termed "sub sectors." These are then combined into 39 "sectors", then into 18 "super sectors", and finally into 10 "industries." The ICB system is used by three major index providers: Dow Jones, FTSE, and Russell.

The competing system is the "Global Industry Classification Standard" (GICS) used by Standard and Poor's and Morgan Stanley Capital International (www.gics.standardandpoors.com). At the lowest level, GICS identifies 147 "sub-industries", which are combined into 67 "industries", then into 24 "industry groups", and finally into 10 "sectors."

At the highest level, ICB and GICS are quite similar; in fact, in eight of ten areas they are pretty much in agreement. These include basic materials, energy, industrials, financial services, telecommunications, technology, healthcare and utilities. Where they disagree is on the proper classification of companies that sell to consumers. ICB distinguishes between companies that sell "Consumer Goods" versus "Consumer Services." Its logic is that, particularly in the U.S. (but increasingly in all developed countries), the latter's revenues have been growing more quickly than the former. In contrast, GICS uses a split between companies that sell "Consumer Staples" and "Consumer Discretionary" items. This logic is based on the observation that latter's sales are more cyclical than the former. However, when you compare the sectors side-by-side, there is a lot of commonality between Consumer Goods and Consumer Staples, and between Consumer Services and Consumer Discretionary. This is similarity also shows up in their returns, which track each other quite closely.

It is also interesting to see these sectors' weight in the overall market portfolio. As the following table shows, sector weights differ across regions:

Sector Weights in Broad Market Index

It is also important to remember that these sector weights are not static; they also differ over time, depending on investors' changing expectations for different sectors' future performance. For example, in 2000, technology had a much higher weight than it does today.

Let us now move on to the question of why one might want to invest in a sector index fund or ETF, instead of the broad market. There are three possible justifications for this. The first is because one's economic position significantly differs from that of the "average" investor who is satisfied to hold the market portfolio. The classic example of this would be someone who holds a large amount of restricted stock in his or her company. This investor might want to ensure that his or her exposure to the domestic equity market did not contain this sector exposure. In this case, his or her portfolio would contain a mix of the other nine sectors.

A second argument that is sometimes made for using sector funds is that an investor might have different risk preferences than the average investor who is content to own a broad market equity index fund. He or she might therefore want to construct a portfolio with sector weights that differed from those in the market portfolio, to achieve a different expected risk and return for his or her equity allocation. This argument is based on the observation than not only have sectors delivered different historical risks and returns, but also that their returns have not been perfectly correlated with each other. This is clearly shown in the following table.

Nominal U.S. Sector Performance 1992-2005

However, we see one significant problem with this argument. In our view, it makes far more sense to calibrate a portfolio to an investor's risk/return preferences by changing allocations across broad asset classes (e.g., by changing the weights given to bonds and equity) than by changing sector weights within just one asset class.

This brings us to the third, and probably the most popular justification for investing in sector (or subsector) index funds: because your forecast for one or more sectors differs from the overall consensus that is implicit in the sectors' weight in the market portfolio. This is active management, pure and simple. And like all other active management, its success depends on superior forecasting skill, which in turn must be based on some combination of superior information or a superior model for deriving insights from it. In turn, these insights must be focused on the two drivers of sector fund returns: either changes in fundamental valuation factors and/or the future behavior of other investors.

The more common name for "the future behavior of other investors" is "trend following" or "momentum." While there are many techniques for analyzing it, the simplest is to invest in the sector with the highest year to date returns, in the expectation that other investors will be doing the same, which will cause its price (and return) to rise still further. For example, the following table shows how different sectors have performed in the U.S. through April 21^st.

Nominal Sector Returns

However, for all its attractions, we all know that this simple momentum strategy also has an important catch: how to accurately forecast when the trend is going to reverse, so you can get out at or close to the top. Unfortunately, history shows that is a lot easier said than done. The great risk of momentum investing is that it drives prices above their fair value, based on the underlying fundamentals of the return generating process. Once that happens, the process becomes very unstable, and prone to sudden and sharp reversal.

This brings us to the question of whether sector index funds are under, over or fully valued today on the basis of their fundamentals. Our basic approach to this issue is to compare the supply of expected equity returns (defined as the current dividend yield plus the expected dividend growth rate) with the return investors should rationally demand (defined as the current yield on real return government bonds plus an appropriate risk premium). Overvaluation is implied when the returns the sector is expected to supply are lower than the returns investors demand (i.e., prices will have to fall to raise dividend yields). Undervaluation is implied when returns supplies are higher than those demanded. And full valuation is implied when they are equal.

Two issues are typically raised about this analysis. The first is the fact that, in addition to dividends, many companies return cash to their investors via share buybacks. To address this issue, we also perform the analysis after adding .5% to the dividend yield to reflect the estimated impact of buybacks. The second issue is the appropriate market equity risk premium to use. While much has been written about this question, there is still no consensus answer (e.g., in "The Equity Risk Premium in January, 2006: Evidence form the Global CFO Outlook Survey", Graham and Harvey found that over a ten year period, the average premium used by practitioners was 3.59%, but with considerable variability). Given this uncertainty, we perform our analysis using both a 3% and 4% equity market risk premium.

Even with these adjustments, it is unfortunately not a straightforward process to apply this analysis to sector funds, because they aggregate so many different companies. This makes it very difficult to estimate an appropriate expected growth rate. Given this, we calculate a "breakeven" growth rate instead, by subtracting the current dividend yield from the sum of the real return bond yield plus an equity market risk premium for the sector. This is the growth rate that implies full valuation of the sector, because it makes the supply of expected returns equal to the returns demanded by investors. If one believes this implied growth rate is too high, the sector is overvalued; if it is too low, the sector is undervalued. The following table presents this analysis for U.S. sector funds, using end of March, 2006 data. The sector risk premia we use reflect the relative variability of the sector compared to the overall market (i.e., the sector beta), calculated over the past three years, multiplied times an overall equity market risk premium. This means that sectors whose returns are less variable than the market require lower risk premia, while those with higher variability require higher risk premia.

In the following table, we show four breakeven real (inflation adjusted) future growth rates for each sector, as well as the overall market, based on different combinations of dividend yield (plus buyback adjustment) and equity market risk premia (3% and 4%).

U.S. Sector Valuations, April 2006

These tables makes two important points.First, in many sectors, the implied breakeven (or fair valuation) real growth rates are quite high relative the implied growth rate for the market as a whole. For example, at a time of unprecedented U.S. current account deficits and consumer borrowing, it is hard to imagine the consumer discretionary sector's dividends forever growing faster than those for the market as a whole. Second, the implied real growth rates for the market as a whole are exceptionally high relative to historical experience, even under the most aggressive assumptions (.5% buyback adjustment and 3% equity risk premium). In the past, dividends growth has been very closely correlated with growth in total factor productivity, which, over the last two years, has only grown by about 2.8% annually in the United States.

In light of these factors, it is hard to escape the conclusion that the U.S. equity market, and many of the sectors within it, are seriously overvalued today. If there are active trades to be made in sector index funds today, it looks like many of them are on the short side. This conclusion is further born out by the growing value of sector ETF shares that have already been sold short (in expectation of future price declines); for example, short interest in consumer discretionary/consumer goods ETFs is now greater than fifty percent of their market value, with energy not far behind with short interest of more than forty percent. Unfortunately, something tells us that not many of these short sales have been made by individual investors.

To be sure, the continued heavy inflows into sector index products suggests that many investors either disagree with this analysis or are basing their investment decisions on their forecast for other investors' future behavior (i.e., momentum) rather than a fundamental value forecast. As we noted above, history has repeatedly shown that this is a very dangerous game to play, and its results are sadly predictable. For example, in their recent paper "The Dumb Money Effect", Lamont and Frazzini found that on average, "retail investors [who switch funds] direct their money to funds which invest in stocks that have low future returns. To achieve high returns, it is best to do the opposite of these investors. We calculate that mutual fund investors experience total returns that are significantly lower due to their reallocations. Therefore, mutual fund investors are "dumb" in the sense that their reallocations reduce their wealth on average. We call this predictability the 'dumb money' effect." Something tells us that when somebody gets around to writing a paper about the results of switching between sector index funds, they will find the same sad results.