The SOA 2015 VBT: How to Read the Mortality Table Behind Life Insurance

The Table Behind the Quote

When a life insurance carrier prices a policy, a life settlement buyer bids on one, or an LE (life expectancy) provider issues an estimate, there is one mortality table sitting underneath nearly all of it: the Society of Actuaries 2015 Valuation Basic Table, or 2015 VBT.

Despite how load-bearing it is, the VBT is poorly explained outside actuarial circles. This post covers what the table is, how to read it, how it relates to the CSO tables and the SSA population tables people more commonly encounter, and what the MP-2021 improvement scale does on top of it.

What the 2015 VBT Is

The Valuation Basic Table series is published by the Society of Actuaries, developed jointly with the American Academy of Actuaries. Each edition is built from mortality experience contributed by U.S. life insurers — the actual deaths observed among people who bought individually underwritten life insurance. The 2015 edition was constructed from roughly 2002–2009 insured-life experience, projected forward to a 2015 baseline.

Three words in the name carry the meaning:

• Valuation: it is the reference table for valuing life-contingent cash flows — pricing, embedded value, experience studies, life settlement analysis.
• Basic: the rates are unloaded best estimates. No conservatism, no profit margin, no padding. This is the actuary's honest answer to "what is the probability this person dies this year?"
• Table: not one table but a family — sex-distinct, smoker-distinct (plus composite), select and ultimate, on both age-nearest-birthday and age-last-birthday bases, covering ages 0 to 120, with a ladder of relative risk variants.

The lineage matters for search: there is a 2001 VBT, a 2008 VBT, and a 2015 VBT. There is no 2021 VBT. If you have seen "2021" attached to mortality work, it is almost certainly MP-2021, the improvement scale discussed below, or the 2017 CSO, the statutory table derived from the same experience.

Reading the Rates

The core quantity in any mortality table is qx — the probability that a person alive at age x dies within one year. Here are actual 2015 VBT ultimate rates (age-last-birthday, non-smoker), expressed both as probabilities and as deaths per 1,000:

Age	Male non-smoker qx	Per 1,000	Female non-smoker qx	Per 1,000
30	0.00046	0.5	0.00034	0.3
40	0.00124	1.2	0.00087	0.9
50	0.00199	2.0	0.00123	1.2
60	0.00428	4.3	0.00316	3.2
70	0.01216	12.2	0.00816	8.2
80	0.04076	40.8	0.02896	29.0
90	0.14417	144.2	0.10638	106.4

Two patterns worth internalizing. First, mortality is astonishingly low through the working years — a 40-year-old male non-smoker who passed underwriting has about a 1-in-800 chance of dying this year — and then compounds roughly exponentially, doubling every seven to nine years of age. Second, the female curve runs consistently below the male curve at every age.

Smoking status moves the whole curve. At age 50, the male smoker ultimate rate is about 4.1 per 1,000 versus 2.0 for a non-smoker — smoking roughly doubles the annual probability of death at that age, which is why it is the single most consequential input on any life insurance application.

Select vs. Ultimate: Why "Years Since Underwriting" Matters

The 2015 VBT is a select and ultimate table with a 25-year select period.

A person who bought a policy last year just passed a medical exam, blood work, and an underwriting review. That screening is information: recently underwritten lives die at materially lower rates than average for their age. The select rates capture this — they vary by both issue age and policy duration. As the years pass, the screening goes stale, and mortality "grades up" toward the ultimate rates, which depend on attained age alone.

This is why a 60-year-old who bought a policy at 59 and a 60-year-old who bought one at 35 carry different expected mortality even at the same attained age. After 25 years, the select effect is fully worn off and everyone sits on the ultimate table.

For general-purpose longevity estimation — where no fresh underwriting has occurred — the ultimate table is the appropriate, conservative choice. That is exactly what our free actuarial life expectancy calculator uses.

The Relative Risk Tables

Real underwriting sorts applicants into classes — Preferred Plus, Preferred, Standard and their many aliases — and each class carries different expected mortality. Rather than leaving carriers to invent class-level rates, the 2015 VBT ships with relative risk (RR) tables: scaled variants from RR50 (roughly half of standard mortality, representing the best preferred risks) up through RR175 (175% of standard).

When a pricing actuary says a preferred-best class "runs at about RR60," they are placing that class on this public ladder. The same vocabulary shows up in life settlement underwriting, where LE providers express an insured's health as a mortality multiplier applied to a VBT baseline.

VBT vs. CSO: Best Estimate vs. Statutory

The confusion between VBT and CSO tables is common enough to address head-on.

The 2017 CSO (Commissioners Standard Ordinary) table is derived from the same underlying insured-life experience as the 2015 VBT, but with margins deliberately added. Regulators require conservatism in statutory reserves, nonforfeiture values, and the definitional limits of IRC §7702, so the CSO overstates expected mortality on purpose.

The rule of thumb: CSO for compliance, VBT for truth. If the question is "what does the law require the carrier to hold?", the answer references the 2017 CSO. If the question is "how long is this person actually expected to live?", the answer starts from the 2015 VBT.

VBT vs. SSA: Insured Lives vs. Everyone

The mortality table most people have actually seen is the Social Security Administration's period life table, which covers the entire U.S. population — insured or not, healthy or not.

Because the VBT reflects only people who passed medical underwriting, its rates sit far below the population table at every working age. At age 50, the SSA 2021 table shows roughly 3.4 deaths per 1,000 males; the 2015 VBT ultimate male non-smoker rate is about 2.0 per 1,000 — roughly 40% lower. At age 70 the relationship persists: about 20.5 per 1,000 (SSA) versus 12.2 (VBT non-smoker).

This is the "select population" effect, and it cuts both ways. If you are insurable, an SSA-table life expectancy likely understates how long you will live — with real consequences for Social Security claiming and retirement planning. If you carry significant health conditions, neither table describes you, which is where individualized adjustment comes in.

MP-2021: Projecting Improvement

Any mortality table is a snapshot, and mortality rates have declined for decades. A table frozen at its baseline year systematically overstates future mortality — and understates life expectancy — for anyone alive today.

The SOA's answer is the MP series of mortality improvement scales, published by its Retirement Plans Experience Committee. MP-2021, released in October 2021, is the final scale in that annual series (the SOA has since moved to its Mortality Improvement Model framework). MP-2021 provides annual rate-reduction factors by age and sex: apply them cumulatively to a base table and the projection reflects expected medical and public-health progress rather than a frozen snapshot.

In practice, "2015 VBT with MP-2021 improvement" is the standard recipe for a forward-looking insured-lives baseline — and it is the published, public foundation under health-adjusted life expectancy modeling.

Where to Get the Tables

Everything above is public and free:

• soa.org hosts the 2015 VBT report, the rate files, and the MP-2021 scale with full documentation.
• mort.soa.org, the SOA's online mortality table database, hosts every variant with a stable table ID. The smoker-distinct ultimate tables on an age-last-birthday basis, for example, are tables 3269 (male non-smoker), 3270 (female non-smoker), 3271 (male smoker), and 3272 (female smoker).

If you want the table's answer without downloading anything, our actuarial life expectancy calculator runs the published 2015 VBT ultimate rates with MP-2021 improvement directly in your browser — no signup, no data sent anywhere.

What the VBT Cannot Tell You

The 2015 VBT knows four things about a person: age, sex, smoking status, and (in select form) years since underwriting. It does not know about diabetes, heart disease, cancer history, kidney function, weight, family history, or any of the other conditions that dominate individual mortality differences.

That is not a defect — it is the table doing its job as a baseline. Individualizing from that baseline requires condition-level modeling: how each diagnosis shifts mortality, how severity grades within a diagnosis, and how conditions interact. That layer is what separates a population estimate from a personal one, and it is the subject of our guide to health-adjusted life expectancy.

Get a personal longevity report to see a health-adjusted estimate built on the 2015 VBT foundation — or if you are an advisor, run one for a client and compare it against the population tables you are using today.