David Blanchett and Thomas Idzorek have a new paper on liability-driven investing (aka LMP, liability-matching portfolios): "LDI Misapplied: Income Portfolios and Liability-Driven Investing
They have (IMHO) three main take aways for those considering an LMP approach to planning their retirement.Liability-driven investing (LDI)—in particular liability-relative optimization—represents a fundamental improvement over more common asset-only portfolio optimization techniques, such as mean-variance optimization. Almost all portfolios exist to pay for some future form of consumption, so liability-relative optimization is almost always more appropriate than asset-only approaches. By considering liability characteristics when solving for the asset allocation, LDI techniques take advantage of the natural hedging quality of certain investments.
Well-meaning, avant-garde practitioners have begun embracing LDI techniques when building portfolios for individual investors (especially retirees) without considering the unique characteristics of the individual’s liability or the risk attributes of the assets or cash flows retirees have available to fund the liability (e.g., Social Security retirement benefits). The result is highly conservative portfolios like those used by institutional investors. But these investors tend to have more predictable liabilities, with idiosyncratic risks leveling out across a large population of retirees.
Individual investors are subject to greater idiosyncratic risk in their retirement spending needs, and must also invest appropriately to hedge risk in an LDI context. Importantly, individuals have greater flexibility in retirement spending, in part because retirement expenses aren’t a legal liability, and because many have other ways to pay for retirement expenses, including other assets, Social Security retirement benefits, and the ability to work during retirement. After investigating better ways to properly model investor expenses, we find that most investors at or near retirement are likely better served with less conservative, more balanced, and more diversified portfolios that recognize the increased duration associated with living longer.
1. Ignoring liabilities (which traditional planning does and LMP tries to fix) has a huge effect on portfolio design.
2. How you model a retiree's liabilities has a huge impact on portfolio design.
3. Ignoring the retiree's full balance sheet (Social Security, home equity, buying an annuity, etc) has a large impact on portfolio design.
Here's a quick look at each of the 3.
Ignoring liabilities results in a very different portfolio than one that accounts for them.
Here's a traditional efficient frontier and here's one that assumes you have a future liability.
Notice that on the right hand side, for the most aggressive investors, things look identical. But as you move to the left, the portfolios begin to look very different. The main point is that unless you are a very aggressive investor, taking liabilities into account will result in a very different portfolio design.
How you model liabilities has a huge impact
Individual investors have a significant disadvantage when it comes to building an LMP: they need to model their future liabilities and they are subject to massive idiosyncratic risk. Insurance companies and pension funds don't have that problem because the law of large numbers means all of the idiosyncratic risk is balanced out.
This is exactly the same reason all of us eschew individual stocks and prefer indexes. But building an LMP for yourself is a bit like investing in a single stock.
Exactly how you model those future expenses will have a large impact on the portfolio you design. The authors show 3 different models of liability and their affect on asset allocation.
With one model, you should have 89% of your portfolio in long (nominal) bonds. With another model only 1.9% in long (nominal) bonds but 18.7% in short term TIPS and 12.9% in global high-yield bonds.
Take the full balance sheet into account
In the real world people have Social Security. They may have a pension. They can buy an annuity if they want. They may have home equity. It has long been known that those things affect asset allocation and that continues to hold true when using a liability-driven approach.
Here is an example of a liability-driven asset allocation that ignores Social Security and another one that assumes Social Security makes up 40% of your retirement income.
Again, on the right side, the most aggressive side, they are identical. But as you become more conservative they diverge.
Security retirement benefits are incorporated into the portfolio optimization the allocation TIPS (both short and long) changes to zero for even the most conservative portfolios. Therefore, while some academics have contended that TIPS are the optimal hedge against the retirement liability, we find that the true benefit of TIPS is likely significantly less when viewed from a more holistic perspective.