Diversification effects in practice
It is undisputed that various asset classes, which in addition to positive return expectations have the lowest possible correlation and increased dispersion among each other, theoretically offer a classic diversification effect. This represents a "free" added value, i.e. a free lunch in the sense of an improved risk/return ratio of the overall portfolio.
In practice, however, things often look different. The following three challenges that limit the scope of free lunch should be mentioned here:
- while the expected returns of the various asset classes are positive, the actual returns achieved can be negative over longer periods of time
- the correlations of most asset classes are usually positive and often higher than assumed in theoretical models
- especially in times of crisis, when portfolios are particularly dependent on the spread of risks, diversification regularly fails
These points lead to the fact that the effective diversification benefits are smaller in practice than can be expected using classical models. Figuratively speaking, most market participants will hardly be satisfied with their tasty free lunch at the end of the day, because it actually contains a lot of empty calories. Therefore, in order to achieve an adequate return, they must ultimately integrate higher risks into their diet. And yet: Diversification across several asset classes remains an overall advantage in most cases, albeit a smaller one than expected.
A question of cost
However, we have neglected one important component in the previous analysis: the fees incurred for ongoing positioning in the individual asset classes. This is the subject of the study "Fees Eat Diversification's Lunch", written by William Jennings and Brian Payne. 1] As the title suggests, the authors argue that the sometimes high costs largely eat away the remaining benefits of diversification.
The study is based on a concept that Charles Ellis already described in a short journal article in 2012. 2] According to this concept, fees should not be measured as a percentage of assets under management, but as a percentage of the actual alpha achieved - and from this perspective, the fees are surprisingly high.
It is known from previous studies that the US equity market - directly or implicitly - is the dominant risk factor for most asset classes and accordingly explains the majority of the returns achieved in institutional portfolios.  A simple way to measure the alpha is therefore to adjust the respective returns for the US equity market factor. Accordingly, the study mentioned at the beginning of this article looks at fees in relation to the alpha of the individual asset classes, which shows their true diversification potential. 
In their research, the authors conclude that the risks of most "diversifying" asset classes can be explained, as suspected, mainly by the beta of the US stock market and have only low alphas. At the same time, the fees are usually significantly higher than for traditional core asset classes. These two effects mean that the true diversification advantage is quite small or - especially in the case of funds of hedge funds - sometimes even disappears completely. In addition, this approach changes the order of the asset classes that come into question for effective diversification.
The authors conclude that the comparatively expensive access to alternative investments in particular represents an economic inefficiency that has remained surprisingly constant over time. The following chart shows the distribution of 45 asset classes sorted by alpha in descending order before and after costs. It should be noted that even positive alphas only have a moderately high probability of actually delivering a positive return contribution in a single year due to usually comparatively high standard deviations, while the fees paid are lost in any case. The considered asset classes are taken from a classification by J. P. Morgan and the used collection of actually negotiated fees by Callan Associates.
In practice, possible diversification advantages must always be weighed against the corresponding implementation costs. These estimates must be considered in combination from the outset in the orientation of a long-term asset allocation. Otherwise it can quickly happen that the well-known economist Milton Friedman is right with his once popular claim:
"There's no such thing as a free lunch".
Experience shows that fees are comparatively high, especially in more exotic asset classes such as hedge funds, private equity and emerging market bonds. If there is any doubt about the diversification benefits remaining after costs, it applies similarly to active vs. passive investments that instruments with low costs are to be preferred.