The official blog of BNP Paribas Asset Management

What makes a ‘good’ factor?

The development of factor investing can of course be traced back to the discovery of factors – although  factor investing aims not just to identify the ‘best factors’, but to exploit them as well as possible.

The development of factor investing can of course be traced back to the discovery of factors – although  factor investing aims not just to identify the ‘best factors’, but to exploit them as well as possible.

Factor investing is based on the notion that stocks sharing common features such as being cheap not only tend to move in sync, but can also generate abnormally high returns over time. Such returns are not well explained by how risky these stocks are, as might perhaps be expected.

In addition to valuation, other common features of stocks known to generate abnormally high returns include highly profitable companies or companies whose share prices rise the most in an uptrend. Furthermore, the volatility of stock prices, a measure of the risk of a stock, seems itself to be a factor in stock performance, with the least volatile stocks generating higher returns than expected.

The other side of the coin is that expensive stocks, shares in unprofitable companies, stocks with the weakest price trends and risky stocks all tend to underperform over time. They typically generate returns well below what should be expected given their level of riskiness.

So there are a number of such factors that can be employed both to describe how stocks in a portfolio move together over time and also to generate positive excess returns over the market capitalisation index. Factor investing is thus about actively controlling the exposure of a portfolio to systematic factors such as value, profitability, low risk or momentum. For this approach to be relevant, the factors themselves must make economic sense and the reasons why they can be used to pick winning stocks must be understood.

The key features of a factor

How can we say that we have identified such a factor?

  • Long-term outperformance: because Financial Theory 1.1 tells us that the risk of a stock should be enough to explain its return, the excess returns of stocks identified with the help of factors are often said to be anomalous and factors are said to be anomalies. But the long-term outperformance of portfolios of such stocks underscores the validity of a factor-driven approach.
  • Be explainable: we will have found a factor when this allows us to provide a rational explanation as to why a given set of stocks offer higher returns than their level of riskiness would imply.
  • Universality: one would expect that a given factor used to pick outperforming stocks in one region of the world should also be effective in a different country or region.

As can be seen in our second video on factor-based investments, in our factor investing strategies, we focus on factors that meet these conditions, namely value, quality, low volatility and momentum.

Are there any others?

The efficacy of a factor is related to its forecasting power when it comes to identifying the stocks that will generate higher returns than implied by their riskiness. But for a factor to be of use for factor investing, it is important that its use does not require an extreme level of portfolio turnover. If it does, transaction costs will quickly exceed any excess return.

An example of such a factor is the short-term reversal factor. It is known that the stocks with the strongest performance in the past week or the past month tend to underperform in the coming week or month. A strategy buying and selling such stocks will, however, require extremely high levels of portfolio turnover rendering it practically impossible to turn a profit.

The capitalisation of a company is another such factor. It is known that smaller capitalisation stocks tend to outperform. But transaction costs and liquidity issues can quickly get in the way of turning a bias towards the smallest capitalisation stocks into a profit, at least for the richer investors.

The issue of the quality of data and hidden biases

Let’s take the example of employee well-being as a factor. A recent study “Linking Workplace Health Promotion Best Practices and Organizational Financial Performance” published in the Journal of Occupational and Environmental Medicine found that stocks of companies in which employees felt the best when it came to health and well-being outperformed stocks of other companies (235% compared to 159% for the S&P 500 index over a six-year period).

This factor is easy to explain: health and well-being = higher productivity. But it raises a number of questions: is the measurement objective and, more importantly, completely unbiased?

Well-being and health are measured via a survey, which means that the findings could be subjective. Moreover, this type of measure is also likely to be biased as well-being will normally vary widely from one sector to another, for example, between tech and industrial companies or between countries with different cultures.

This example illustrates that there are other conditions for using a factor:

  • It must be based on objective and high-quality data with a long enough history to validate its long-term efficacy in picking outperforming stocks.
  • It should not have hidden structural biases that might be the cause of most of the observed performance.

As the amount of information generated by companies continues to grow, other factors may become relevant. But as we will see in our next article, it is not about quantity, it is about how factors are used that makes one approach more relevant than another .

Related articles

Weekly insights, straight to your inbox

A round-up of this week's key economic and market trends, and insights on what to expect going forward.

Please enter a valid email
Please check the boxes below to subscribe