Comparing the Approaches to Liability-Relative Asset Allocation:

Shopping for a gift isn't a one-size-fits-all approach, and neither is asset allocation when managing a liability. Just like each individual can want something different, each liability-relative asset allocation strategy can vary depending on different factors. What do you think is the primary factor that determines which liability-relative approach to use?

No, actually. The frequency of cash flows also is a crucial part of the process.

Right. Just like gift giving depends on the person receiving the gift, the liability-relative approach used in asset allocation depends on the clients and their specific needs. So you wouldn't want to use the hedging/return-seeking portfolio for a bank because its liabilities would be changing, nor for an underfunded pension plan.

Not really. That's an important factor, but the time horizon also plays a role.

In practice, the basic approach is the surplus optimization because it's basically the traditional asset-only mean–variance model. It works best over a single period and can handle any funded ratio and all risk levels. Plus, since the surplus optimization's inputs are expected returns and a covariance matrix, the assets and liabilities are linked through correlation conditions. This leads to a straightforward correlation between assets and liabilities. What kind of correlation do the assets and liabilities have in a surplus optimization?

Indeed! The surplus optimization leads to a linear correlation between the assets and liabilities because of its straightforward usage of a covariance matrix and expected returns. So it's really a simple approach.

Not quite. That would mean that assets would be increasing as liabilities decrease.

That's not it. There's one covariance matrix that links assets and liabilities.

Simplicity is also a characteristic of the hedging/return-seeking approach, which creates two buckets: a liability hedge and a return-seeking portfolio. This approach is best for risk-averse investors, especially ones that have a surplus of assets, but it can be used for all risk levels. If there's no surplus, then only a variant of this approach can be used. The hedging portfolio can be built using a variety of different strategies, including one that focuses on the underlying drivers of the liability. What approach do you think that is?

No. Optimization is focused on maximizing utility, not drivers of the liability.

Yes. Factor models serve to understand the underlying drivers of the liability, so that's a perfect way to hedge the liability portfolio and also link the assets to the liability. Speaking of linkages, the integrated asset/liability portfolio is technically a linked portfolio between the assets and liability, and it too can be created using the factor model.

That's not it. That's not an asset allocation model approach.

This direct link between assets and liabilities allows you to model transaction costs, turnover constraints, and any other real-world constraints, which makes this approach more dynamic than the others. What's another way to describe the integrated asset/liability approach?

Not so. With the ability to model multiple constraints, this approach isn't that straightforward.

No. This approach is more dynamic than a routine asset allocation.

Exactly! The integrated asset/liability approach is a complex asset allocation decision that's based on the link between assets and liabilities, so as the liability changes, the assets will adapt and change as well. That means this approach is perfect over multiple periods. It can function on a linear or non-linear correlation between the two portfolios and at any risk level.

To summarize: [[summary]]

The time horizon

The client's needs

The frequency of needed cash flows

Linear

Inverse

Non-normal distribution

Optimization

Factor models

Yield curve analysis

Linear

Routine

Complex

Continue

Continue

Continue

Continue