The dynamisation of asset liability management is explored by Marcus Burkert and Thorsten Kaspar, who show how the allocation and management rules are optimised simultaneously

Asset liability management (ALM) denotes the mutual matching of assets and
liabilities with the target of an integrated overall control. Often, simulation methods (eg, Monte Carlo simulation) which develop a multitude of possible future development paths and subsequently analyse their impact on crucial target variables are used for this purpose.

ALM evolved out of the separate consideration of assets and liabilities (sequential ALM). However, this approach neglects the fact that parameters from both sides of the balance sheet can have reciprocal effects on each other. Consequently, the next step was the so-called simultaneous ALM that takes account of these interactions. The latest development is dynamic ALM that assumes that the asset allocation is not fixed but can develop dynamically over time depending on the available risk budget.

In practice, hedging concepts such as CPPI (constant portfolio protection insurance) and overlay management are often considered in ALM studies. The systematic of the hedging concepts is taken as a fixed management rule for the simulation. Setting these rules beforehand can significantly change the results. It makes sense to elaborate the different effects of management rule variations. The complexity of a model increases exponentially with the number of alternatives, so the rules that are to be changed and optimised should be selected carefully.

The dynamic ALM approach intends to optimise not only the asset allocation, but also the management rules at the same time. The result is not merely a recommendation for the ideal asset allocation, but also for the design of the rule system. The following example is to clarify this approach.

A pension fund intends to optimise its asset allocation with the help of an ALM study. Besides optimisation criteria (eg, performance, build-up of reserves, etc) the passing of a stress test is crucial. The stress test allows for an examination of the extent to which the assets can cover the liabilities of the pension fund even under different extreme market conditions. Value deductions are made especially for the highly market sensitive asset classes (eg, equities). These investments, which are called risk assets, influence the passing of the stress test to the largest extent. The stress test is only passed, if the reduced assets are still sufficient for covering the liabilities.

In the example study, the assets of the pension fund should therefore be aligned in such a way that the risk assets are maximally allocated with a proportion that still allows for passing the test at the end of any future business year, but at the same time at least to an extent for still having perceivable influence on the overall return.

A fixed proportion of risk assets can hardly meet the goal of an annual passing of the stress test. Market price variations will inevitably occur over time. The total assets will be reduced due to falling prices and the coverage of the liabilities could be endangered. As a consequence, a strict adherence to a fixed strategic asset allocation is not possible if the test is not passed.

This is taken care of through the dynamisation of the ALM projection method. For this purpose, a reduction of risk assets due to a failure to pass the test is considered as a dynamic management rule.

In a further step, the dynamic management rule can be extended in such a way that the strategic risk asset quota is not only reduced and restored path-dependently, but is rendered dynamic itself. The current assets, liabilities and the liability forecast for each year lead to the respective available risk budget from which the maximum proportion of risk assets can be deduced. The purpose of the ALM analysis is therefore - besides the effects of a dynamic risk asset quota - to identify which proportion of the risk budget should be spent in a business year - 100%, or would 80% or even 60% be more reasonable for ideal results?

The example illustrates a possible increase of the risk quota in year one, which is lowered again in the following years. At the same time, the results have proved that a complete usage of the risk budget increases the probability of passing the test in contrast to a static adherence to the strategic asset allocation. Compared with a lesser usage of the respective risk budget, however, it does not seem ideal. Yet too little usage of the risk budget is not ideal either, because it leads to a reduction of the target achievement probability in the long run without further decreasing the capital investment risks in return.

The analyses are based on different underlying target allocations that differ in their proportions of non-risky assets and are modified through a dynamic risk asset quota. A dynamisation also leads to differences between the individual basic allocations. Some basic allocations prove to be more advantageous when modifying the degree of risk budget usage. Thus, they can generate a higher added value than other allocations when rendered dynamic.

A comparison of the crucial key figures for different target allocations that are subdued to modified dynamic management rules allow for the identification of the ideal allocation and management rule pairs.

Overall, the analysis technique allows for a profound insight into the effects of a dynamisation of capital investments and risk budget parameters on the target parameters of the pension fund. Besides the ideal allocation an ideal management rule can also be defined with its help. Together, they allow the requirements concerning the asset side induced by the liabilities to be met.

In addition to the illustrated example, any target parameter can be integrated into a management rule and optimised.

Marcus Burkert is managing director at HEUBECK-FERI Pension Asset Consulting; Thorsten Kaspar is senior consultant at Feri Institutional Advisors
sed simultaneously