The Nobel prize-winning work of Robert Shiller has been used by acolytes to try to parse indices around the world © FT montage; Reuters; Bloomberg

The last decade in financial markets could plausibly be called the Age of Indexation. The rise of passive investment has seen hundreds of billions of dollars move into index funds, which has coincided with a growing belief that the index can be used as the principal building block in investing.

A corollary of this has been the near-universal acceptance that indices can be studied over time to provide both broad and specific conclusions about what investors should do with their money. Robert Shiller, the Yale professor, won a Nobel prize for his theory based on top-down valuation analysis of indices to judge future stock market returns. Acolytes have taken his work and applied it in some form to professional money management. Modern finance is addicted to explaining the world through indices.

For everyday savers indices continue to have practical value as a simple way to gain exposure to equities and bonds. For most people they probably remain the worst form of investing apart from all those other forms that have been tried. The problems begin when intelligent observers use historical observations about indices to draw sweeping conclusions about the world today that may be far less meaningful than they believe.

Index-based analysis in its crudest form takes a common valuation yardstick, such as the price to earnings ratio, and compares the value of the index today on this basis to its historical average. Its fundamental logic is one of mean reversion: if the current valuation of the index on any chosen metric is above its historical average then “stocks” are “expensive” due to a tendency for the market to revert.

There are many popular versions of this thinking, with arguably the most popular based on Professor Shiller’s “cyclically adjusted” price to earnings ratio that attempts to smooth the “e” component over an economic cycle. (Debate over this version’s specific merits continues to rage, and there is not space to go into it here. However as a brief aside, is it really meaningful to cyclically adjust back Facebook’s earnings from 2008 to 2018, or is it just an analytical nonsense? Should we cyclically adjust back motorcar sales from 1930 to 1920?)


Proportion of listed US stocks in 1900 that was in rail. Much of today’s broad equity value is accounted for by technology

A big issue with using indices in this way is not with mean reversion as a concept, but that the data set it uses to derive its mean has changed so frequently that it may not be producing a meaningful average to revert back to.

Modern indices are to a large extent still constructed on a country-by-country basis. Investors wanting to invest in US stocks will pick the S&P 500, those keen on Spanish stocks will choose the Ibex 35 and so on. In the past this made sense as most of their components did the majority of their business in their domestic markets. This is no longer the case, and it raises problems when reaching grand conclusions about the market today based on the index in the past.

Many important S&P 500 members derive an increasing amount of their sales outside the US. As such, charts that used to try to measure the ratio of the index to US gross domestic product over time make less sense. Apple, the largest component of the index, sells more outside the US than within. Is it an American company, or a global one? Similarly, companies in Spain’s Ibex make just 34 per cent of sales at home, and 46 per cent outside the EU. Does the historical average accurately reflect this shift, and how should we account for it?

A different problem with index-based analysis is that it glosses over huge changes in the component parts of these benchmarks over time. This becomes particularly problematic for those who rely on historical average valuation multiples for an index to draw conclusions about today. Elroy Dimson, Paul Marsh and Mike Staunton of the London Business School have shown how in 1900 over 60 per cent of the value of listed US equities was in rail. Today much of this value is accounted for by technology, which came into existence as a category relatively recently and has undergone huge evolution in the past decade.

At the same time returns on invested capital for the most profitable US-listed companies have exploded since the turn of the millennium. According to McKinsey, ROIC, excluding goodwill, for non-financial companies in the 90th percentile of profitability surged to above 80 per cent by the financial crisis, and further since. This compares to less than 25 per cent for this cohort in the 1970s.

What this means is that those who put average valuations from the past next to those of today are comparing hugely capital intensive operations of the past to vastly more profitable businesses that logically trade on higher multiples than the railroad and steel companies that were dominant more than a century ago. Again, how do we adjust for this when looking at historical comparisons? Is it even meaningful to do so?

The world is a constantly changing. Looking at the average does not tell the real story; frequently this is taking place at the margins. While it may be reassuring to boil this uncertainty and complexity down into simple measurments and formulas this is likely to only ever provide a superficial imitation of reality.

Get alerts on Passive Investing when a new story is published

Copyright The Financial Times Limited 2019. All rights reserved.
Reuse this content (opens in new window)

Follow the topics in this article