Has it really been 10 years since I authored the epic “Variable Annuity Guaranteed Minimum Death Benefit” – The Movie, in Contingencies Magazine? [I don’t care if you
don’t remember, my mom considers it an epic, so it’s an epic.] And what prophecy – so many of us have in fact vacated the offices mentioned therein: “There are two kinds of actuaries: those whose careers will end due to the GMDB, and those who’ll move into their offices.” And to prove timing is everything with comedy, that article was listed as “Humor”, but a few years later, I guess it wasn’t so funny.

So, ten years later, what is new with GMDB?  Well, consider that only a few people have  been involved in VA guarantees over the last decade, and I’m one of them. More  important, they asked me to write this article. So, now the reader will tolerate my views on three aspects of GMDB – the retail product, the reinsurance players, and the reinsurance techniques. By reading this article, you will be revered at actuarial cocktail parties, and I shouldn’t have to tell you how important that is.

Retail Product

In general, a 2005 GMDB risk profile (i.e. what you stand to gain versus what you standto lose) is much more favorable than the 1995 risk profile. Back then, the GMDB was the cutting edge for product developers, and great effort was made to differentiate the variable annuity through the GMDB. The time period was marked by constant
incremental design changes, as companies attempted to improve upon their competitor’s
offerings. Usually, these GMDB changes were modest, such as a 7-year ratchet changing
to a 6-year ratchet, but given that fifty or more companies were playing this ‘incremental’
game, over time GMDB design became much more aggressive.

A more aggressive GMDB design for the risk taker meant a more valuable GMDB design
for the retail customer, which in itself is not bad for business. But, the insurance industry
too often gives things away for free, a common practice with the GMDB circa 1995.
Back then, product design efforts resulted in an ever richer GMDB being offered in the
VA chassis, at little or no cost to the retail customer. Marketers claimed that it was
difficult to charge for something that had no value. There were a few actuarial voices in
the wilderness proclaiming the risks of GMDB, but they were either shouted down or
dispatched with offers of new computers and/or free ice cream. But then the stock
market crashed…

Although the market crash caused pain for writers of variable annuities, it also changed
the mindset for VA guarantees. Finally, the GMDB was recognized as a valuable benefit,
and this paved the way for VA writers to charge a more reasonable fee for the coverage.
More than any other reason, this is why today’s GMDB risk profile is far superior to a
decade earlier. There are other reasons, for sure, such as more balanced investing and
tighter control of benefit options, but make no mistake, collecting 40 basis points instead
of 10 creates a much rosier picture for the GMDB.

Today, some VA writers are even assessing the GMDB fee on the guaranteed benefit,
rather than the account value. Of course, this still doesn’t preclude losses, but it does
ensure that the VA writer collects meaningful fee income in all scenarios, even when the
account value swoons. This approach also aligns the VA writer with the reinsurance
market, where players have always stressed the “catastrophic” nature of the risk, and the
importance of being paid commensurate with potential losses.

Players

It may surprise observers but I don’t ever remember a period where reinsurance was not
available for GMDB! If this was a biblical epic, we would say that Transamerica begat
Swiss Re and CIGNA Re, Swiss Re begat AXA Re, and AXA Re begat ACE Tempest
Life Re. [God must have taken CIGNA Re before it could begat.] But these are just
some of the companies that have publicly professed interest in GMDB business. Beyond
these names were almost all other reinsurers! These other reinsurers assumed GMDB
business at one time or another, either as a concession to a client, through a retro
arrangement, or perhaps to test the waters.
To illustrate, my company has existed for 7 years, and we’ve assisted clients with dozens
of treaties. By my count, we have used six different companies to assume GMDB
business, and ACE Tempest is the only one of the six that has publicly stated that they
reinsure GMDB! So, why the confusion on supply of GMDB reinsurance? Well, here’s
one reason, but I warn you, it may hurt: If you’re a reinsurer that claims to be a great
partner, and an expert on risk, it must be a little embarrassing to not have a solution for
GMDB. What better way to avoid your client’s scrutiny than claiming [loudly] that
nobody offers GMDB reinsurance. But as I can attest, when nobody’s looking, many of
the reinsurers are entering into one-off GMDB treaties.
Interesting of late, there has been an emergence of new reinsurers for GMDB, who are
not shy about professing their interest in the business. Of course, this new group of
reinsurers will manage their risks like an investment bank, and actually, in many cases
the reinsurers are owned by investment banks. Generally, for these companies, GMDB
risk will not be assumed unless it can be hedged through the capital markets. This is
quite different from reinsurance that was traditionally offered, but this new “banking”
model should still be quite useful to ceding companies wishing to divest GMDB risk.
One big challenge to the banking model is how to support new business. In the past,
reinsurers provided a 2-3 year window within which the terms of the reinsurance treaty
were set for all business sold during the window. This approach matched the interests of
the VA writer, where products have to be filed with the SEC and states, and where it is
very difficult to gain shelf space for a product that changes frequently. By securing
reinsurance terms, the VA writer could confidently market its product for a year or two
without concern for the cost of managing their risk. But the capital market approach
depends on derivatives, where prices change continually, and it is outside the comfort
zone of bankers to have pricing commitments that extend far into the future. Here’s my
advice to the bankers — focus on inforce blocks. It is far too time consuming to get the
VA writers to change their thinking toward daily pricing for annuities, so selling a treaty
covering new business will always be problematic. Inforce blocks that could benefit
from reinsurance are still plentiful, and besides avoiding the issue of future derivative
prices, the demographics and investment allocation of inforce blocks is known.

Reinsurance Product

Compared to ten years ago, today’s traditional reinsurance treaties have a great deal of
risk sharing between the ceding company and reinsurer. I’m often asked, “Tim, can’t I
get one of those GMDB deals that were available in 1995?” I’m also asked, “Tim, why
do you dress so poorly?” but that answer has nothing to do with this article. Anyway, to
the former question the answer is no. This answer evokes a sigh of dismay, as if no buyer
could resist a 1995-style program. The truth is that most buyers did resist those programs
in 1995! While the reinsurers of 1995 wrote plenty of treaties, I can attest that they lost
more than they won. And though many treaties were lost to other reinsurers, even more
were lost because the ceding company decided to self insure. Ultimately, most of these
self insurers regretted not divesting of their risk management through reinsurance.

Time will tell whether reinsurance buyers will look back fondly and wish they could
secure one of those 2005-style programs! As I noted, today’s traditional treaties tend to
have more risk sharing, but make no mistake, the risk transfer in these treaties is huge.
For example, a treaty could have a stop loss deductible of 15 bps, with the reinsurer
effectively on the hook for all claims in excess of 15 bps. Obviously, this is risk sharing,
but the ceding company has secured a cap on their losses. On an expected basis, the
ceding company will pay the bulk of the claims, since most scenarios will not create
claims in excess of the deductible. But when you ignore expected results, the risk
transfer in this arrangement is significant, since in those few scenarios where claims
exceed the deductible, all the excess is transferred to the reinsurer.

Similarly, a reinsurance treaty could have an aggregate claim cap, which is a risk sharing
element. But usually, at least in the treaties my company is involved with, this claim cap
is based on the very worst scenarios for account value performance and mortality. In
theory, actual claims could exceed the claim cap. But even if this occurs, it only means
that the reinsurer has paid claims far in excess of their premium collections.

Besides the introduction of more risk sharing, today’s traditional reinsurance treaties
offer better profit potential to the reinsurers; that is, the reinsurance premium rates are
materially higher than 10 years ago. Back then, a 1-year ratchet may have had a
reinsurance premium of 15 bps. But today, the retail fee for the ratchet may be 40 bps,
and the reinsurer may get 35 bps, based on the treaty’s allocation of risk. With GMDB,
it’s natural to focus on claims, but premiums are also important, and revenue per unit of
risk has greatly increased over the last ten years. As evidence, a leading GMDB reinsurer
reported earnings of $26 million in the first quarter of 2005! [Public documents do not
provide a split of the $26 million by product reinsured.]

Speaking of product lines, I’ve found it curious that the companies that provide GMDB
reinsurance are not more adamant about securing other forms of reinsurance. For
example, if I provided GMDB to a client, why not also request life reinsurance? This is
not a tie-in sale, but instead a request by the reinsurer to match the life reinsurance offers
from other reinsurers. In the long run, this seems like a win-win, since the reinsurer is
able to diversify and grow their business, and the ceding company broadens their
relationship with a reinsurer that has greater sophistication and risk management
capabilities than a life-only reinsurer.

The bank model reinsurance treaty is relatively new, but given its dependence on actively
traded derivatives, there are some features that one can expect:
First, as mentioned earlier, there will be a tendency to apply the bank model to
inforce blocks rather than new business.
Second, treaty terms will be firmer for the treaty’s first decade than for later years.
The key building block for risk management will be SPX options on the exchange, and
for now, these options are probably only useful for 10-12 years into the future. Given
that the bank model requires near complete hedging, it might be unusual for the ceding
company to lock in treaty terms for the life of the contracts.
Third, risk sharing will still occur with the bank model, because the reinsurer will
not want to accept risks related to lapses, asset allocation, and in some cases mortality.
Fourth, prices on new reinsurance blocks will move with shifts in the price for
derivatives, so that interest rates and option volatility become very important in managing
expectations for future reinsurance premiums.
Fifth, it will be relatively easy to secure reinsurance coverage at the extreme tail
(e.g. SPX goes to zero), because these outcomes are covered by derivatives. But much
like the traditional reinsurance market, there is a price for everything and covering the
theoretical tail will be more expensive.

OK, so that’s the summary of the last ten years for GMDB. Most of the best ideas for
reinsurance of GMDB date back a decade, but it took ten years for some of these ideas to
move forward. In the interim, companies have come and gone, offices have been
vacated, and stochastic modeling is mainstream. Actuarial work is little different from
other work, in that sometimes it takes a crisis before necessary changes can be
implemented.

Comments are closed