Investor-Defined Liability Curves

This paper proposes a new model to describe the asset-liability structure of investors with intangible, strategic wealth “goals.” Goals in this case are distinguished from tangible liabilities—such as corporate bonds—which have known future values. By incorporating additional stochastic parameters and utilizing a classifier algorithm to determine what is and is not a “goal,” this paper attempts to show that any contemplated liability structure can be modeled quantitatively by combing the new model for intangible liabilities with existing ALM models for tangible liabilities—which are already well developed. This is particularly important for certain investor classes—such as endowments, foundations and individual investors—where strategic goals tend to make up a very large or often a predominant percentage of their total liability structure. It can also be utilized to a lesser extent in more complex ALM applications—such as with insurance companies, banks and pensions—to help fill gaps in areas where intangible parameters are still present (such as defining shareholder return on equity targets). Similar approaches have been taken with Goals-Based Investing (GBI) frameworks. However, the GBI approach in this paper seeks a closer alignment with LDI frameworks so that it can be integrated alongside existing LDI and mean-variance (MPT) approaches . The primary distinguishing features of this models versus others is that 1) risk aversion is an output based on spending goals, not an abstract input based on heuristic categories and 2) the future liabilities themselves are stochastic (similar to assets) and are positively correlated with assets, which allows more flexibility and presumably more precision in modeling them under mean-variance and LDI frameworks. VSQ believes that modeling goals as peaked probability density functions also aligns more closely with Prospect Theory versus other approaches.While the proposed model hopefully serves to advance theoretical work in this field, the model presented is also intended as a practical and implementable model for use in working directly with investor clients—albeit with significant parameter tuning and advisor engagement required