In this article (What Advisors Need to Know About Annuity Mortality Credits) at advisorperspectives.com, Joe Tomlinson lays out his explanation of Mortality Credits and how to use them in decision making around annuities. He also makes a case that the environment for annuities isn't as dire as past research makes it sound. I don't know anything about that. The last thing I read was about annuities, SDP, and optimal annuitization age by G. Irlam where he says that if one is making an all or nothing wager on annuities then sometime before 80 makes sense and probably earlier if legging in. That and there are combinations of age and wealth that will make it more optimal to annuitize than other combinations. I'm guessing that whether Joe is right or not in this article, and I'm sure he is, age 59 is just a little bit too early any way you cut it unless one were to be hyper-risk-averse.
Here I just wanted to make sure I understood Joe's math and methods for my future use. I tested it out on me. Here are what I took to be the steps:
1. Look at the SOA mortality table. I couldn't make sense of the longevity extension adjustments they (SOA) do but the basic table I get. I can transform that table for a given age and it's probably close enough without the adjustment for this post's purpose. For a 59 year old the mean expected value is age 85.9. The median is around 86 and a half and the 95th percentile is over 99. We are using the mean later for pricing a bond with similar duration. Note that the SOA data assumes a healthier pool because of selection bias among other things. I suspect there is a little bit of an extra conservatising gouge in there too since the main users are more often than not insurance companies...but maybe that's too close to conspiracy theory territory for today.
2. Price an annuity. I used AAcalc.com to estimate a payout for a 59 year old Floridian and a money's worth ratio of .95. Using a corp bond assumption I come up with $5734 for a payout on $100k. I validated that using immediateannuities.com and had at least one quote for a simple life annuity of 5712. Close; let's use aacalc 5734. I don't have access to CANNEX data.
3. Weight the annuity price by survival probabilities. Take the percent remaining alive estimate in the 59+ age vector and multiply by the annuity payout. Combine that with the 100k purchase and perform an excel IRR calc. I do this and come up with 3.07%. Recall that the industry term Payout would be ~5.73% which can be considered, depending on how you look at it, meaningless.
4. For some year calculate the 1 year survival probability. For a 59-60 step in this scenario that is 99.58%.
5. Determine the mortality adjusted return (MAR) for that same year. Take the IRR and divide by step 4 so: (1+.0307)/.9958 - 1 = 3.50%
6. Figure the mortality credit at age. Mortality adj return minus the IRR: .0350-.0307 = .0044 mortality credit (rounding effects).
7. Price a bond for similar duration. The diff between mean mortality and 59 is close to 27 years so let's look at the 30 year treasury which was yielding ~.0287 today. Add a little fudge for a spread (Joe uses 1%+). So lets call a yield estimate at ~.0387 or more.
8. Compare the bond and the MAR. bond = .0387 MAR = .0350 so the difference is -.37%
9. Decide about the annuity. If the result, which includes a little fuzz, is zero or negative, that is supposed to mean no-go for the annuity decision. The result? No go. Maybe next year...or age 75.
No comments:
Post a Comment