Consider how your factor choices affect your risk premia
Increasing numbers of investors are embracing factor investing. Many lower cost vehicles, designed to give exposure to popular styles, have been launched to meet the demand. But do investors really know what they’re getting? There is a dizzying array of outcomes, even for seemingly well-defined strategies.
Take ‘value’ for example. It’s probably the oldest ‘style’ of the lot, dating back at least to the days of Benjamin Graham. The earliest style indexes included value based offerings, and most modern multifactor ETF’s and other quantitative processes include some measures of value in their mix. But even in an apparently well-defined style such as this, there can be hugely different results, contingent on the choice of factors and how they’re implemented.
Chart 1 shows the relative performance of ten different value factors in the US over the last 35 years. In each case, cap-weighted sector-neutral portfolios are formed monthly based on the top quintile of the factor. The obvious correlation between the different factors is plain to see: all factors were hit during both the dotcom boom and the global financial crisis. But it’s also clear that there’s a big disparity between the factor premia. It matters which value factor or factors are used. A strategy based on buying high cashflow/price shares fared considerably better than one based on book/price. Understanding how a fund is exposed to value at the factor level is vital as it uncovers the type of value a manager is targeting. In turn, this helps to understand better the processes underlying the strategy which leads to more precise differentiation and classification of funds.
Value factors are amongst the most correlated. There is much greater dispersion in other popular style categories, such as quality, volatility and momentum. Portfolio analysis at the factor level shines a light that uncovers exposures that would otherwise go unseen.
Portfolio construction matters: one factor, many outcomes
It’s not just the choice of factors that matters – implementation can also make a huge difference. Suppose you were looking to build a simple value ETF, and had narrowed down the desired factor exposure to choosing high cashflow/price stocks. The choice of construction and rebalancing rules has big repercussions too.
Chart 2 shows the range of risk and return results for different concentration, weighting, rebalancing and sector positioning choices. The different colors correspond to various rebalancing frequencies, ranging from monthly to annually. The shapes represent differing weighting schemes: market cap, equal or fundamentally weighted. The marker sizes relate to the concentration levels of the portfolios: whether comprised of the top half, quintile or decile of cashflow stocks. And finally, the labelling indicates whether or not the portfolios are sector neutral.
There are seventy-two different combinations in this simple analysis. In reality, there are many more approaches, incorporating a wide array of risk targeting methods and implementation rules. But even in these stylized examples, the spread of outcomes (across the same 35-year window) is enormous. Ex post tracking errors range from 3%pa to over 13%pa, and annualised outperformance from 60 basis points pa to more than 3.5%pa. And all this with the exact same factor: high cashflow/price.
Understanding what’s in a fund from its description, say ‘value’, does not do justice to the range of possibilities and nuances that can be present. To truly understand what’s in a fund, detailed analysis at the factor level provides far greater insights in confirming a manager’s strategy and unmasking hidden tilts.