We are in an industry that relies on assumptions. Implicitly we make them every time we invest our client’s money.
Some are simple – do we believe that active management adds value, or do we assume not and so prefer passive for our investments? Similarly, we generally all believe – as I wrote last week – that adding asset classes and sub-asset classes improves diversification and so delivers a superior risk/return to clients, but the extent to which this is the case varies, probably more than we typically assume. If we do not regularly challenge these assumptions, we risk delivering rather naively-constructed portfolios – ones that might have made sense in the past but, with time and greater understanding, no longer continue to do so.
Let us think through some examples. First, we generally all assume that equities will be one of the higher return asset classes in the long term (compared, for example, to bonds or cash). Is this fair? Well, the past decade or so tells us not, and certainly during many periods in history sub-asset classes of fixed income (for example, high yield) have substantially outperformed global equities. But while we all probably do expect equities to outperform bonds in aggregate in the long term, the margin between them is in all likelihood not something we would agree on. Similarly, and even more importantly, comes the sub-asset class level assumption. We typically invest a proportion of assets in equities to get a higher return, and as a result expect higher risk. But where in equities do we invest? Do we think emerging markets are higher risk and higher return? Probably. But what about developed equities: are they all the same in the long term? And how do we access them? We rely upon using indices for modelling purposes as proxies for asset classes and markets, but since investing is not costless, do our assumptions hold?
All too often we make the implicit assumption that the returns and risk of indices are something we can directly invest in. They are not. For example, the average equity manager has substantially more invested in mid and small cap stocks than his index. Any asset allocation based on the index is therefore likely to deliver different ‘real world’ results for both risk and return. Second, should we include assumptions about outperformance (alpha) when building an asset allocation for a client that will be invested actively? And then what about beta? All other things being equal, a lower beta manager will bias down the absolute volatility of a portfolio and so any optimisation including this manager will under-allocate if it assumes index volatility potentially disadvantaging a client in a falling market. The reverse is also true, meaning a client’s portfolio could take too much risk if using higher beta active managers possibly to the investor’s advantage in an up market but not so in a falling market. Such changes in market direction are difficult to predict and are often not part of long-term modelling assumptions. The assumption that indices give us enough data from which to make decisions is hence frequently either wrong or only half the story.
