Asset-Liability Management (ALM)

This work is a simplified version of the process described in two scholarly articles that Travis Jones, PhD (Florida Gulf Coast University) and I wrote. (An Application of Asset-Liability Management for Financial Planners by Brown and Jones, Journal of Financial Planning, 2011. and Integrating Asset-Liability Risk Management with Portfolio Optimization for Individual Investors. by Jones and Brown, Journal of Wealth Management, 2009.)

The notion is rather simple. Suppose an investor wants to ensure that funds will be available for future spending. Much like a pension plan or insurance company, an investor can estimate the amount of funds needed in the future. Once these portfolio liabilities are identified, we can purchase low-risk assets with durations (and amounts) that offset the timing (and size) of the liabilities.

In my opinion, a rather practical purpose of using such an approach is to help an investor determine the minimum fixed-income allocation. For example, ALM might suggest that an investor has at least X% in high-quality fixed-income investments given a Y level of portfolio withdraws (timing and amounts withstanding)

in addition to a given immunization period.

One way to think of an immunization period might be the notion that within this period risky-non-immunizing assets might under-perform safer immunizing assets (think bonds beating stocks from 1999 to 2009)

hence immunizing the risk that your investments are a detriment to your needs. For example, if you believe the risk of bonds beating stocks is less than, say, 5% over the next 7 years, then 7 years might be a good immunization period. Others may argue that an immunization period should be longer or shorter for other reasons entirely.

Also, consider interest rates. Interest rates play a large role in solving for the ALM portfolio. When interest rates are high, it takes less bonds to generate and offset your liability needs. When rates are low (as in 2012), it takes more bonds to immunize your liabilities. So given the low rate environment, does ALM argue for overweighting low-yielding investments? Not necessarily. Consider an investor who has a high degree of certainty that bonds will under-perform over the next 5-years. In this case, a shorter immunization period would be set. On the other hand, ALM is largely a risk management tool, so setting a conservative immunization period is prudent. As a result, balancing the realities of the capital markets with conservative risk managment is key.

Finally, consider that this is one representation of an asset-liability model. In our model, we make a simple assumption that a spending rate today (along with inflation) is a fair representation of future portfolio demands. A small tweak to the model and hard dollar cash flows can be modeled.

I have seen a variety of models that have a much different flavor. Another approach, used mainly by acedemic professionals I believe, rely on the notion that future outflows are stochastic, dynamic, essentially less predictible. As a result, their modeling of future spending involves inductive reasoning models, probabilistic decision trees, and other techniques.

Aside from how future cash flows (liabilities) are modeled, I have seen different approaches to identifying the best assets to offset liabilities. Say we’ve mapped out a liability, we take the simple notion that it has a duration of 7 years, for example. So an immunizing bond equivalant would also have a duration of 7-years

simple enough. So putting together a portfolio of immunizating bonds to offset all modeled liabilities is pretty straight forward. Other approaches, however, might look at a cluster of liabilities and then attempt to postulate that a blend of XYZ securities (assets) will offset this sub-group of liabilities. This exercise is repeated for all liability clusters.

Which is the best approach? From a private wealth perspective, I personally do not believe the more complicated approaches are any better or worse. What I like about our approach is the strong logic, simplicity, and the use of less assumptions (both for modeling liabilities and matching assets).