Signal From Noise

February 22, 2011

Signal From Noise

Broad market funds may still reign supreme among equities exchange-traded funds, but in the past few years, sector funds have increasingly colonized the equities space. Today, sector ETFs span dozens of categories and billions of dollars in assets, accounting for nearly 17 percent of all investment dollars in equity ETFs.

It’s easy to see why. Sector funds allow investors to fine-tune their equity exposure, and convert insights and opinions about macroeconomic factors into actual investment allocations. Whether investors seek to capture returns from outperforming industries, or protect their portfolios from the wrong end of the business cycle, sector funds provide the tools to find and exploit the signals present in the noise of the broad market.

But dividing stocks into individual sectors isn’t as easy as it might first appear; one man’s technology company can be another’s alternative energy play. Seemingly minor discrepancies in classification methodology can result in major differences in indexes based on sector classifications, and their patterns of returns.

In this paper, we examine the three major classification systems for sector indexes—ICB, GICS and TRBC—in order to determine the most effective one for ETF construction. We analyze these benchmarking systems along three vectors: the size and depth of their securities universes; the structure of their sorting methodologies; and inter-sector correlation analysis. We conclude that the systems are more similar than different, but that ultimately one classification system—TRBC—stands out as the most useful for ETF investors.

Three major classification systems dominate sector indexing worldwide:

  • The S&P/MSCI Barra Global Industry Classification Standard (GICS)
  • The ThomsonReuters Business Classification (TRBC)
  • The Dow Jones/FTSE Industry Classification Benchmark (ICB)

Although the details differ, the goal of each is the same: to sift through tens of thousands of available stocks and organize them into discrete yet comprehensive categories. A sound classification system will aggregate companies in a way that not only makes intuitive sense, but produces low-correlation investable buckets.

To evaluate which of these systems best suits the ETF investor, we must take a top-down approach. Our criteria are simple: To be effective, a sector benchmark should a) track a large portion of its relevant stock universe; b) make logical sector definitions; and c) categorize businesses in a rules-based way.

Thus we base our analysis of these systems on three simple, distinct criteria:

  • The size/depth of their securities universes
  • The structure and relevance of their sorting methodologies
  • The correlations between individual sectors

Establishing The Universe: Size Matters
Thousands of companies fill the world’s equity exchanges, but not all are relevant to an ETF investor; some are too small, others too illiquid and so on.

Therefore, all classification systems must first select some universe of securities from the greater global equity market to organize. This universe sets caps on the number and market capitalization of equities available to populate individual sectors.

Out of the three classification systems, TRBC assembles the broadest starting universe of securities, giving it the most potential components with which to populate its sectors. (See Figure 1.)


Figure 1


Find your next ETF