Limits of Diversification in a Multi-Factor Context

• The correlation between different factor exposures can at times be significantly negative.
• Due to these dependency structures, offsetting effects may arise in a multi-factor context.
• Well-constructed multi-factor strategies, such as Value Size and Quality Momentum, show positive diversification effects.

Dependency Structures Between Different Factors

There are limits to what is possible when it comes to the efficient construction of multi-factor portfolios. Not all factors can be meaningfully combined. In practice, supposed multi-factor portfolios often show a significant exposure to only one or two of the targeted factors. In addition, unwanted side effects may arise, meaning that the expected diversification benefits fail to materialise. The reason for this is that dependency structures between factors are not sufficiently taken into account in the portfolio construction process.

Table 1 shows the linear correlation between different factor exposures. While, for example, a value exposure is positively correlated with dividend and size, it is negatively correlated with profitability, leverage and also momentum. In other words, attractively valued companies tend, on average, to have an above-average dividend yield and a comparatively low market capitalisation. At the same time, however, they tend to show below-average profitability and elevated leverage. Quality companies, by contrast, are characterised by above-average profitability and below-average leverage — features that are, conversely, associated with higher valuations.

These dependency structures have a direct impact on portfolio construction. In the case of so-called pure factor strategies, such as a pure value strategy, it is possible to achieve neutral exposure relative to the global equity market despite the negative correlation with quality exposures. Although minimal exposure to the value factor is sacrificed, the gain in terms of neutralising quality exposures outweighs this effect.

Table 1: Correlations of Selected Factor Exposures

Effective Implementation of Multi-Factor Strategies

In multi-factor strategies, combinations of positively correlated factors are generally suitable, as emphasising one exposure automatically leads to a stronger exposure to the other. This allows the greatest possible factor exposure to be achieved. For example, a Value Size strategy delivers greater value exposure than a pure value strategy, in which the size exposure must be kept neutral despite the positive correlation. It can therefore be concluded that positively correlated factors can be implemented particularly efficiently in a multi-factor strategy, for example compared with a separate 50/50 weighting of the two individual factors.

But what about negatively correlated factors? Here, there are fundamental limits in terms of the maximum factor exposure that can be achieved, regardless of the holistic portfolio construction process. For example, part of the value exposure is inevitably lost if a positive momentum exposure is to be achieved at the same time. In addition, in the case of negatively correlated factors, joint optimisation of portfolio construction often provides no substantial advantage over a simple 50/50 combination of the respective individual strategies.

For maximum flexibility, the following rules of thumb can be derived for the construction of factor portfolios:

1. Positively correlated factors should be combined within a joint multi-factor strategy in order to achieve the greatest possible factor exposure.
2. Negatively correlated factors, by contrast, can be combined effectively on a separate basis: there are no significant losses in terms of factor exposure, while diversification across factors and greater flexibility in allocation are preserved.

Assenagon Equity Framework

This does not mean, however, that investors necessarily have to choose between the two approaches — either the efficient combination of positively correlated factors or the use of negatively correlated factors for diversification. This is precisely where the Assenagon Equity Framework comes in: it makes it possible to combine different factors flexibly and thereby not only maximise factor exposure, but also realise attractive diversification benefits.

The following analysis examines the diversification of different return drivers using the factors value, size, quality and momentum. First, two portfolios are formed from the respective positively correlated factors Value & Size and Quality & Momentum. Figure 1 shows the active rolling one-year excess return of the two multi-factor portfolios versus the global equity market.

Figure 1: Rolling Active One-Year Returns

Over the entire observation period since 2000, the average return is positive at around 2.5% for Value & Size and 1.4% for Quality & Momentum. Both approaches deliver long-term added value versus the global equity market on a standalone basis. What is more, the returns of the two strategies are almost uncorrelated. With a correlation close to zero, attractive diversification effects arise, supporting the simultaneous use of both strategies.

Figure 2: Frequency Distribution of Rolling Active One-Year Returns