Generating optimal portfolios within the FTK framework

The analysis of an optimal portfolio is normally handled by classic mean variance quadratic programming. This analysis forms the basis of ‘asset only’ investing, and can be converted simply for tracking error investing (relevant to fully invested asset managers). The problem requires generalisation to include liabilities. Once this is done, we find not only the residual of asset-only investing, but also additional terms representing the influence of the liabilities. The optimal portfolios are thus different from classic mean variance optimisation, and moreover, the nature of the portfolios changes depending on the funding status of the ALM problem. Thus, ‘asset only’ mean variance optimisers produce inferior portfolios, and inappropriate recommendations for the ALM problem. Further, while we recognise the historical status awarded to volatility as the risk measure, we note the ability of the formalism to handle non-standard risk statements, such as the new financial assessment framework (FTK), recently proposed in the Netherlands. The results can be summarised as a new form of capital efficient frontier, not in return-volatility space, but asset return-shock space and a new tool that characterises the trade-off between competing volatility and shock constraints.