The Effects of Decision Flexibility in the Hierarchical Investment Decision Process

Large institutional investors allocate their funds over a number of classes (e.g. equity, fixed income and real estate), various geographical regions and different industries. In practice, these allocation decisions are usually made in a hierarchical (top-down), consecutive way. At the higher decision level, the allocation is made on basis of benchmark portfolios (indexes). Such indexes are then set as targets for the lower levels. For example, at the top level the allocation decision is made on the basis of asset class benchmark indexes, on the second level the decisions are made on the basis of sector benchmark indexes, etc. Obviously, the lower levels have considerable flexibility to deviate from these targets. That is the reason why targets often come with limits on the maximally allowed deviation (or quot;tracking errorquot;) from these targets. The potential consequences of deviations from the benchmark portfolios have received very little attention in the literature. In this paper, we discuss and illustrate this influence. The lower level tracking errors with respect to the benchmark indexes propagate to the top level. As a result the risk-return characteristics of the actual aggregate portfolio will be different from those of the initial benchmark-based portfolio. We illustrate this effect for a two level process to allocate funds over individual US stocks and sectors. We show that the benchmark allocation approaches used in practice yield inferior solutions when compared to a non-hierarchical approach where full information about individual lower level investment opportunities is available. Our results reveal that even small deviations from the benchmark portfolios can cause large shifts in the top-level risk-return space. This implies that the incorporation of lower level information in the initial top-level decision process will lead to a different (possibly better) allocation