Reimagining index funds

Financial indices started proliferating a century ago, but were long little more than badly-constructed, infrequently-updated ways to measure economic vim long before the birth of GDP. In 1957 that changed forever.

That year, Standard & Poor’s launched a market capitalisation-weighted index of the biggest US companies, as opposed to the older, but smaller and price-weighted Dow Jones Industrial Average. From then on, indices would reflect what the actual “market” was doing.

The index was also calculated daily (and soon every five minutes) by a “Datatron” computer that was hooked up to Wall Street’s stock market ticker machines. Robert Shiller argues this was “the beginning of the electronic era in finance”.

Almost 70 years later, the S&P 500 remains the gold standard of stock market indices — with tens of trillions of dollars tied directly to its performance — and the cap-weighted approach is the dominant way to construct all financial benchmarks.

OK, enough throat-clearing. Why this detour into the history of financial indices? Because Research Affiliates’ Rob Arnott reckons benchmarks like the S&P 500 can be improved to make better index funds. From a paper just put out by RA:

Publishers of commercially available cap-weighted indices typically limit the number of names in an index to improve investability, adding and deleting stocks from the index over time. As a result, these indices are not purely passive. Additions are usually growth stocks, trading at premium multiples and with impressive momentum, whereas most deletions — unless occurring because of a merger or acquisition — have the opposite characteristics. 2 Some index additions go on to great success, exceeding all expectations, while others do not. Some deletions proceed to stock market oblivion, and others handily recover. We even see flip-flops: stocks added then quickly dropped from the index, and vice versa. Collectively, additions and deletions encompass most of the turnover in a cap-weighted index. Although small in magnitude, some of this turnover can be harmful to our wealth. Buying mainly frothy stocks with strong momentum and selling mainly tumbling stocks that are severely out of favor, creates an unhelpful buy-high and sell-low dynamic in a cap-weighted index.

Improving on the classic index approach of simply weighting according to a company’s stock market capitalisation (or how much debt it has outstanding) is not a new idea.

These days, indices are mostly weighted by their free-float market cap, for example. The S&P 500 itself is actually selected by an index committee from a pool of eligible stocks. And over the years academics have decomposed equity market returns into various “factors”, which practitioners have used to build investment strategies that supposedly improve on good old plain beta. This is basically RA’s bread and butter.

The company’s flagship suite of indices — born out of a 2004 paper titled Fundamental Indexation — weight companies according to underlying measures like revenues and profitability rather than their market value. Research Affiliates manages about $141bn through various systematic strategies tied to its indices.

Now Arnott and his colleagues Chris Brightman, Xi Liu and Que Nguyen think they’re on to a new winner by in effect fusing the fundamental approach of its RAFI methodology with traditional cap-weighting.

What if we no longer chase soaring winners and abandon tumbling losers, and instead choose stocks based on a more stable metric, namely, the size of the underlying business? Selecting a stock based on the size of its macroeconomic footprint rather than on the market’s expectation of the company’s future success comes very close to the same portfolio held by existing cap-weighted indices. Big businesses are usually large-cap, and small businesses are usually small-cap. Thus, if we select stocks based on the economic scale of the underlying business rather than on market-cap, the result is a portfolio with superb liquidity and capacity, fully comparable to popular and commercially available market-cap indices.

Essentially, its “Research Affiliates Capitalization-Weighted Indices” use the fundamental data methodology to filter through the investment universe, but then uses market capitalisation to weight them as normal.

The result, Research Affiliates says, is a suite of indices that look, smell and talk similar to the likes of the S&P 500, the Russell 1000 or the MSCI All-World Index, but actually walk a little better.

According to simulated backtests on 30 years of data (and notably including live prices since September 2021) its version of the S&P 500 outperformed the original by 12.6 per cent. A Research Affiliates version of the Russell 1000 outperformed by 15.3 per cent. Not huge amounts over a long period of time, but enough to be meaningful.

Here is a chart showing it versus the relative dollar growth of the RACWI 500, RACWI 1000, Russell 1000 and a mechanistic “pure” cap-weighted version of the S&P 500 against the real S&P 500 since July 1991.

Why the slightly different performance? A separate RA paper soon to be published in the Financial Analysts Journal goes into more depth, but basically it seems to boil down to the RACWI indices excluding some high-flying but small companies and including big but unfashionable ones.

This, RA argues, helps their indices costly index “flip-flops” that drag on performance, through minimising trading costs, avoiding getting sucked into overhyped index inclusions and prematurely dumping unsexy but still big index deletions

Markets have periods of comparative tumult and comparative stability. The performance difference (or tracking error) between any two cap-weighted indices with different constituent stocks, and their respective ability to add value, should vary, roughly, in parallel. If markets are perfectly efficient, the choice of market capitalization or fundamentals to select constituents would be irrelevant because market-cap should correctly value a company’s future business prospects. If markets are inefficient, pricing errors will exist. Assuming the pricing errors mean revert, then selecting index constituents based on their fundamentals should add value by largely avoiding the buy-high/sell-low dynamics inherent in selecting index constituent stocks by market-cap. During turbulent times, when mean reverting pricing errors are presumably larger, RACWI should exhibit higher tracking error against conventional cap-weighted indices and capture greater incremental returns as these outsized pricing errors mean revert. This is exactly what we see in the historical data.

The effect can be chunky. Arnott et al estimate that stocks ejected from the S&P 500 actually outperformed new entrants by an average of 2,200 basis points in the year after the index rejig (FTAV has written about the index inclusion effect before). Or as Arnott told FTAV in an email:

. . . It’s easy to see how to “beat the market”: identify some class of investors who are content to lose. For both RAFI and RACWI, success is funded by those who chase the frothiest fads, buying after the market has fallen in love with an asset, and selling assets that are feared and loathed.

The RACWI indices have remarkably modest tracking error to their conventional cousins, but is there a hidden “factor tilt” that can explain the outperformance — such as how greater exposure to small caps explains why equal-weighted indices tend to do better over time?

Yes, to an extent, but it’s not huge, the investment house argues. Unsurprisingly, there is “small, but enough to matter” tilt towards value, but it actually “beats value relentlessly”, says Brightman. The RACWI 500 index actually has a negative size factor, ie it includes more big stocks than the S&P 500 (which typically includes about 120 stocks that aren’t actually one of the 500 biggest stocks in the US by market cap). It also has a slight loading of momentum.

Will this “reimagine index funds” though, as Research Affiliates hopes?

FTAV gathers that they are already in talks with some intrigued investors, but we’re sceptical that this will become a breakthrough hit. Human inertia is a more powerful force than gravity, and traditional cap-weighted indices and index funds are entrenched as the standard. There’s a reason why equal-weighting — the New Coke of index funds — hasn’t taken off either.

But given the subtle flaws of classic cap-weighting indices — especially if implemented unthinkingly for index funds, as opposed to just being pure benchmarks — it is great to cool new research and people trying to do things differently. Even if some of them turn out to be New Coke.