OIAD Working Paper 2023-01
Agency decision
Ask Donna
What actually matters in this document.
Text
OIAD Working Paper 2023-01
September 2023
Investor Testing Report on
Registered Index-Linked
Annuities
KATHERINE CARMAN, Senior Financial Economist, Office of the Investor Advocate, Securities and
Exchange Commission, 100 F Street, NE, Washington, DC 20549, carmanka@sec.gov, ORCID ID:
0000-0002-5760-4441
ALYCIA CHIN, Senior Financial Economist, Office of the Investor Advocate, Securities and Exchange
Commission, 100 F Street, NE, Washington, DC 20549, readlinga@sec.gov, ORCID ID: 0000-00029570-0549
JONATHAN COOK, Financial Economist, Office of the Investor Advocate, Securities and Exchange
Commission, 100 F Street, NE, Washington, DC 20549, cookjon@sec.gov, ORCID ID: 0000-00016067-0960
STEVEN B. NASH, Data Analyst, NORC at the University of Chicago, 4350 East-West Hwy 8th Floor,
Bethesda, MD 20814, nashst@sec.gov
BRIAN SCHOLL (Principal Investigator; Corresponding Author), Chief Economist, Office of the Investor
Advocate, Securities and Exchange Commission, 100 F Street, NE, Washington, DC 20549,
schollb@sec.gov, ORCID ID: 0000-0001-5088-6952
DAVID B. ZIMMERMAN, Financial Economist, Office of the Investor Advocate, Securities and Exchange
Commission, 100 F Street, NE, Washington, DC 20549, zimmermanda@sec.gov, ORCID ID: 00000001-5784-2733
ACKNOWLEDGEMENTS
We thank staff at Fors Marsh for helping us conduct this research, particularly Alexsandra Galanis, Emily
Appel-Newby, Daniel Greenberg, Ashley Simpkins, Susan Ullrich, and Hunter Peebles. We also thank staff of
the SEC, particularly members of the Office of the Investor Advocate (OIAD) and Division of Investment
Management, for helpful comments on our research design and background on legal requirements. Steven Nash
is an on-site institutional contractor to OIAD.
On December 29, 2022, Congress enacted the Consolidated Appropriations Act, 2023 (Appropriations Act).1 In the Appropriations Act, Congress
indicated that “the Commission shall prepare and finalize, new or amended rules, as appropriate, to establish a new form . . . on which an issuer of
a registered index-linked annuity may register that registered index-linked annuity, subject to conditions the Commission determines appropriate .
. . with respect to the registered index-linked annuity.”2 In addition, Congress indicated that, in designing the form, “the Commission shall . . .
engage in investor testing; and incorporate the results of the testing required . . . in the design of the form, with the goal of ensuring that key
information is conveyed in terms that a purchaser is able to understand.”3 The investor testing discussed in this report was conducted in response
to the Appropriations Act.
DISCLAIMER: The report was prepared by staff of the Office of the Investor Advocate at the U.S. Securities and Exchange Commission (SEC)
but does not necessarily reflect the views of the Commission, the Commissioners, or other members of the staff. The results of simulation
analysis contained in this report should not be construed to predict future performance of any registered index-linked annuities (RILA) product or
promote any RILA.
See U.S. House of Representatives, Consolidated Appropriations Act, H.R. 2617, Pub. Law No: 117-328 (Dec. 29, 2022), available at
https://www.congress.gov/bill/117th-congress/house-bill/2617.
2
See id.
3
See id.
Investor Testing Report on Registered Index-Linked Annuities | 1
1
Table of Contents
1. Introduction and Executive Summary .........................................................................................4
2. RILAs: Structure of Contracts and Investment Options ..............................................................6
Indexes ........................................................................................................................................ 6
Investment Terms ........................................................................................................................ 7
Value Calculations ...................................................................................................................... 7
Insurance Features ....................................................................................................................... 7
Charges, Penalties, and Fees ....................................................................................................... 9
3. Overview of the RILA Market and Simulated Performance During Historical Periods ...........10
Data ........................................................................................................................................... 11
Methodology ............................................................................................................................. 11
Limitations ..............................................................................................................................12
Descriptive Statistics on Overall RILA Market ........................................................................ 13
RILA Indexes, Investment Terms, and Insurance Features ...................................................... 13
Illustrative Examples Comparing Investment Option Value to an Index ................................. 17
RILA Performance During Various Market Conditions ........................................................... 19
Comparing RILA Features ........................................................................................................ 22
Variations in Insurance Features and Simulated Returns .......................................................22
Variations in Term Length and Simulated Returns ................................................................26
Summary Performance Statistics .............................................................................................. 29
Risk............................................................................................................................................ 30
Summary and Discussion .......................................................................................................... 32
4. Review of RILA Marketing .......................................................................................................34
5. Qualitative Testing .....................................................................................................................35
Methods ..................................................................................................................................... 35
Limitations ..............................................................................................................................36
Results From Round 1 ............................................................................................................... 36
Results From Round 2 ............................................................................................................... 37
Summary of Qualitative Testing ............................................................................................... 39
6. Quantitative Testing ...................................................................................................................41
Methods ..................................................................................................................................... 41
Sample ....................................................................................................................................41
Study Design and Overview ...................................................................................................43
Investor Testing Report on Registered Index-Linked Annuities | 2
Comprehension Measures.......................................................................................................44
Moderating Characteristics .....................................................................................................45
Limitations ..............................................................................................................................45
Results ....................................................................................................................................46
Testing Impacts.......................................................................................................................47
Subgroup Analysis ..................................................................................................................52
Time ........................................................................................................................................53
Summary of Quantitative Testing ............................................................................................. 55
7. Conclusions ................................................................................................................................57
Summary of Methods ................................................................................................................ 57
Summary of Findings and Discussion....................................................................................... 58
Implications of the Research: The Economics of RILAs .......................................................... 60
Implications of the Research: Considerations for Disclosure Design ....................................... 62
Concluding Thoughts ................................................................................................................ 63
References ......................................................................................................................................64
Appendix A. Supplementary Analyses on the RILA Market ........................................................67
Appendix B. Materials Used in Round 1 of Qualitative Testing ...................................................68
KIT Disclosure .......................................................................................................................... 68
Scenario Text............................................................................................................................. 70
Appendix C. Materials Used in Round 2 of Qualitative Testing ...................................................71
Introduction to RILAs ............................................................................................................... 71
KIT Document Tested: Important Information You Should Consider About the Contract ..... 72
Graphics and Table Shown in Testing ...................................................................................... 73
Scenario Text............................................................................................................................. 76
Appendix D. Materials Used in Quantitative Testing....................................................................77
Introductions Presented in Quantitative Testing ....................................................................... 77
Introducing Participants to Key Terminology ........................................................................77
Marketing Materials That Could Be Presented to Consumers When Shopping for a RILA..77
Plain Language Introduction to Factors That Could Influence Participants’ Decision to
Purchase a RILA .....................................................................................................................77
Versions of the KIT That Were Shown in Quantitative Testing ............................................... 78
Investor Testing Report on Registered Index-Linked Annuities | 3
1. Introduction and Executive Summary
Registered index-linked annuity (RILAs) are retirement investment products geared
towards retail investors that have at times been suggested to offer higher growth potential than
fixed-income instruments, combined with a degree of risk protection. Entering into a RILA
agreement means that a retirement-oriented retail investor will have an annuity contract with an
insurance company. Investors generally purchase RILA products through a financial
professional. Despite the retail-focus, RILAs are arguably complex products that may be difficult
for a retail investor to evaluate or understand and that typically require investors to make
complex ongoing financial decisions, thereby raising interest in investor protection concerns.
The benefits of investments in RILAs depend on exactly how these contracts are constructed and
used, with investors having to choose (among other things): an insurance provider, the
underlying linked index, the insurance features that may limit some risks and potential upsides,
and the “investment term” – the period over which investment returns are calculated. These
factors significantly affect the degree to which a RILA product may be an appropriate investment
for a particular retail investor. RILAs typically have implicit fees and can also be associated
with significant charges and penalties, including those assessed for withdrawals from the
contract, withdrawals that occur in the middle of an investment term, and withdrawals before age
59½ (due to tax penalties).
In 2022, Congress directed the U.S. Securities and Exchange Commission (SEC) to
“establish a new form . . . [to] register[] index-linked annuity” products. 1 Congress mandated
that the form must be designed “in a manner to ensure that a purchaser using the form receives
the information necessary to make knowledgeable decisions, taking into account (1) the
availability of information; (2) the knowledge and sophistication of that class of purchasers;
(3) the complexity of the RILA; and (4) any other factor the Commission determines
appropriate.” 2 Furthermore, “the Commission shall . . . engage in investor testing; and
incorporate the results of the testing required . . . in the design of the form, with the goal of
ensuring that key information is conveyed in terms that a purchaser is able to understand.” 3
These activities were to be completed so that “not later than 18 months after the date of
enactment of [the] Act, the Commission shall prepare and finalize[] new or amended rules, as
appropriate . . .” 4
This report presents the results of the research that fulfills the 2022 Congressional
directive to “engage in investor testing.” In this report, we discuss the research methods we used
to examine RILAs and address issues raised by the 2022 directive. First, we examined the RILA
market and how RILAs operate, providing information on the complexity of these products and
the consequences of certain investor decisions. Additionally, we summarized the products
1
See U.S. House of Representatives, Consolidated Appropriations Act, H.R. 2617, Pub. Law No: 117-328 (Dec. 29, 2022), available at
https://www.congress.gov/bill/117th-congress/house-bill/2617.
2
See id.
3
See id.
4
See id.
Investor Testing Report on Registered Index-Linked Annuities | 4
available on the market as of early 2023, thus providing an up-to-date view of RILA contract
offerings. We used simulations to examine how specific features, such as the limits on upside
and downside risk, may affect investor outcomes. This analysis illustrates the complexity of the
products, the ways particular features work, and the importance of investor decisions about
RILAs.
Following our market analysis, we engaged in qualitative and quantitative testing for a
draft sample Key Information Table (KIT), a hypothetical disclosure that contains information
about RILA contracts and their features and risks. 5 In our qualitative testing, consistent with
best practices from communications design (see, e.g., Morgan et al., 2002), we aimed to
understand investor responses to the KIT and to generate hypotheses about areas of confusion
and impediments to investor understanding. Qualitative testing is critical to this project to better
understand the barriers that everyday investors can experience in attempting to comprehend
financial information. At the same time, because qualitative testing occurred with a limited
sample of participants, it is essential for follow-up research to examine the KIT communications
with a larger sample. As such, we also conducted quantitative testing with approximately 2,500
participants to examine specific design features of the KIT.
Throughout the research, our primary questions of interest were:
1. How do different features of RILAs affect investment outcomes in relation to these
products?
2. What are investors’ reactions to RILA disclosures? What areas of understanding and
misunderstanding arise when viewing the KIT?
3. Does background information about a RILA affect understanding of key concepts about
RILAs? Specifically, what background information engenders the highest levels of
understanding as a lead-in to viewing the KIT: (1) information about the contract
purpose, (2) a plain-language introduction to a RILA, (3) a benefits-only introduction, or
(4) no additional information?
4. Does the design of the KIT affect consumers’ understanding of RILAs? Specifically,
does a question-and-answer (Q&A) format lead to higher comprehension than a
statement-based format?
5. How does comprehension vary across investors with different levels of investment
sophistication?
Ultimately, findings from these efforts provide insight regarding the extent to which investors
have the information they need to make knowledgeable decisions about RILAs, and they suggest
that there is potential for misunderstanding that could affect consumer welfare. We conclude by
providing implications of our work for future research and policymaking.
In this report, all references to the KIT refer to different hypothetical disclosures used for testing purposes. There may be modifications in the
disclosure requirements that the Commission may propose for RILAs to fulfill the 2022 congressional directive.
5
Investor Testing Report on Registered Index-Linked Annuities | 5
2. RILAs: Structure of Contracts and Investment Options
Insurance companies offer RILA contracts, each of which is accompanied by a menu of
investment options. 6 Over the life of a RILA contract, the investor picks from this set of
investment options, making decisions about the underlying index, the “investment term” (i.e., the
period during which investment returns are calculated), and the corresponding insurance
features. Each time an investment is made (including at the end of an investment term), the
investor can choose to allocate funds to one or more investment options. Throughout this report,
we use the word “contract” to refer to the larger product that may include multiple investment
options selected over the course of ownership. Investment options are typically characterized by
an index; investment term; charges, fees, and penalties; and insurance features.
Importantly, certain decision-relevant attributes pertain to the overarching RILA contract
(e.g., surrender charge period), whereas other attributes pertain to investment options (e.g., limits
on upside and downside risks and investment term). The two-part structure, consisting of both
the RILA contract and the investment options, adds a layer of complexity to RILA products,
which may make investment decisions about RILAs difficult for potential investors and
purchasers to understand.
RILAs generally have two phases: an accumulation phase and a payout phase. During
the accumulation phase, the investor makes contributions to the RILA, chooses investment
options, and accumulates returns through the investments. During the payout phase, the investor
receives income from the RILA. Income may take the form of a lump-sum payment or, in some
cases, the investor may choose to convert the lump sum to a stream of future income (an
annuity).
Indexes
The value of an individual investment option depends, in part, on the performance of the
underlying index. 7 As noted below, certain indexes are relatively broad based (e.g., S&P 500),
whereas others are more specialized (e.g., SPDR Gold Shares [GLD]). Certain insurance
companies also use proprietary indexes. It is reasonable to assume that continued growth in the
RILA market will lead to an expansion of the number of indexes used.
A RILA can be offered either as a standalone product or as an option within an annuity contract that also offers other options (e.g., variable
annuities).
7
An underlying index could refer to any index, rate, or benchmark (such as an exchange-traded fund or product that tracks an index) used in the
calculation of positive or negative interest credited to an investment option.
6
Investor Testing Report on Registered Index-Linked Annuities | 6
Investment Terms
Each investment option is associated with an “investment term.” 8 The investment term
determines the time period during which changes in value of the investment option are
calculated. In other words, gains or losses are calculated at the end of each term, and the
associated value of the investment changes. As noted, the investment term is associated only
with investment options, not the RILA contract overall (which could last longer). This
difference means that the investor may select several investment options (each of which has its
own investment term) over the life of the RILA contract.
The role of investment terms also creates a situation that may be unique for RILA
purchasers compared to other investments they hold. In particular, RILA investors periodically
realize gains or losses at the end of each investment term: essentially forcing liquidation (sale of
the investment option) at a specified date, even if market conditions are unfavorable. In contrast,
during a down market, a mutual fund investor, for example, could wait to sell the fund, avoiding
realizing those losses. Thus, the investment term feature adds a “timing risk” for RILA investors
compared to certain other investments.
Value Calculations
There are two common methods for calculating the change in value for an investment
term: the point-to-point method and the average method. For the point-to-point method, the
change in value is calculated by comparing the value at the start of the investment term to the
value at the end of the investment term. Various calculations, such as adjustments due to the
insurance features, are then applied to this value calculation. For the average method, the change
in value is calculated based on periodic evaluations of the index level within an investment term
and then averages those periodic evaluations. Similarly, credits and deductions are then applied
based on this average performance.
Insurance Features
RILAs are distinctive because they offer insurance features that limit the potential losses
of an investment relative to the underlying index. However, the insurance features also limit the
potential gains. The insurance features apply only to the investment option, not to the RILA
contract itself, and do not provide protections against penalties that might be applied for early
withdrawals or other reasons.
8
In Registration for Index-Linked Annuities; Amendments to Form N-4 for Index-Linked and Variable Annuities, Investment Company Act
Release No. 33-11250; 34-98624; IC-35028 (Sept. 29, 2023) (the "proposed rule"), the investment term is referred to as the “crediting period.”
However, the phrase investment term was used in the qualitative and quantitative testing. As a result, we use the phrase investment term
throughout this report.
Investor Testing Report on Registered Index-Linked Annuities | 7
There are two insurance features that can limit the losses for an investment option:
•
•
Floors: A floor is a maximum loss (percentage) on the investment option. For example,
a 10% floor will protect the investor from any loss greater than 10%, but the investor
bears any losses up to 10%. A 5% floor would provide greater protection than a 20%
floor.
Buffers: A buffer absorbs losses on the investment option up to a certain point. It does
not turn losses into gains. For example, with a 20% buffer, the insurance company
absorbs any losses up to 20%, but the investor bears any additional losses. In other
words, if the index decreased by 25%, a 20% buffer would reduce the loss to 5%.
Alternatively, if the index decreased by 15%, a 20% buffer would reduce the loss to zero.
A larger buffer provides greater protection.
In general, a buffer protects the investor from experiencing small losses for the investment
option, but it does not protect them completely against large losses. In contrast, a floor protects
an investor from experiencing large losses but does not protect against small losses.
Insurance features can also limit gains for an investment option:
•
•
Caps: A cap is a maximum gain (percentage) on the investment option. It reduces the
potential gains from investing in the option compared with investing in the components
of the index to which it is linked (such as through an investment in an index mutual
fund).
Triggers: A trigger fixes gains to a specific rate so long as the index’s gain is over a
corresponding trigger “threshold”. For example, consider an 8% trigger with a trigger
threshold of 0%. The RILA would return 8% as long as the index experienced any
positive return (i.e., above 0%), including above 8%. Thus, a trigger is more valuable
when index returns are between the threshold and the trigger percentage (in this case,
between 0% and 8%). As a result, a trigger below the index returns during the
investment period is likely to limit gains relative to the index, whereas a high trigger
could boost gains above the returns experienced by the index.
In general, both caps and triggers limit returns to RILA investment options relative to large
increases in the value of the underlying index. However, a trigger can provide higher returns
than the underlying index, when index returns are low.
A final insurance feature may act on both gains and losses for an investment option, but
may also act on gains or losses alone:
•
Participation rates: The participation rate acts as a multiplier on index performance.
For example, a two-sided 85% participation rate reduces both gains and losses to 85% of
Investor Testing Report on Registered Index-Linked Annuities | 8
the gains or losses on the underlying index. For example, a gain of 10% on the index
would be reduced to a gain of 8.5% on the RILA and a loss of 10% on the index would
be reduced to a loss of 8.5% on the RILA. A participation rate can be over 100%, in
effect creating a leveraged investment product. An investment option with a participation
rate may also have caps or other limits on gains.
Charges, Penalties, and Fees
Both RILA contracts and investment options can be associated with significant, avoidable
financial penalties for investors. For the purposes of this report, we concentrated on three
financial penalties:
•
•
•
Surrender charges: A RILA contract can define a “surrender charge period” and
charges for withdrawing money from the contract during that period. These charges
occur even when the initial investment option’s term is shorter than the surrender period.
Interim Value Adjustment and mid-term withdrawals: These penalties occur when
money is withdrawn from an investment option before the end of its term (e.g., when
money is moved out of an investment option to another option, even if the funds are not
removed from the contract). 9 An interim value adjustment may be positive or negative.
It has the potential to reduce the value of the investment option. In some situations, the
investor could lose up to 90% of their investment.
Tax penalties: In addition to other tax implications, penalties can occur when the
investor withdraws money from the RILA contract before age 59½.
To avoid all these penalties, an investor would need to hold the investment throughout the full
surrender period, hold the investment until after the investor turns 59 ½, and withdraw money
only at the end of each investment option’s investment term. As a result, to avoid charges and
penalties, an investor must likely select and manage a RILA contract for several years, making
decisions about several investment options over the life of the contract.
Annual fees are charged in some RILA investments, but the practice is not as prevalent as
annual fees in the mutual fund industry, for example. As an alternative to charging annual fees,
providers often make money in other ways such as by earning more on their investments than
they promise to pay investors (Moenig, 2021).
RILAs measure the change in index values between the first and last day of the investment term. If you withdraw money from an investment
option before the end of the term, the annuity contract specifies how the value of the investment option is determined. In some contracts, the
value will be a pro-rata portion of market performance, whereas other contracts will include complex formulas based on interest rates and
derivatives pricing models. Losses can even exceed any floor or buffer that would be available if an investor held the investment option until the
end of the term.
9
Investor Testing Report on Registered Index-Linked Annuities | 9
3. Overview of the RILA Market and Simulated Performance During
Historical Periods
In this section, we provide an overview and examination of the RILA market, including
descriptive statistics on RILA contracts and investment options, and how RILA investment
options with different insurance features would have performed using simulations of various
historical time periods. RILA products are complex, and the implications of their insurance
features are not obvious. In particular, insurance features affect gains and losses depending on
future index returns over the investment term. Unfortunately, at the time the investment is made,
investors will not know these future returns. Additionally, insurance features are correlated with
each other (as shown below), so a certain level of downside protection also limits possible gains.
For an investor considering a RILA product, natural questions include: Do RILAs offer
higher returns than the underlying index over time? Does the protection of buffers or floors
outweigh the foregone gains from caps? How do the insurance protections protect my
investments? Although part of the answer to these questions relates to an investor’s preferences,
particularly their sensitivity to risk, those preferences can only be considered well informed if the
implications of the RILA contract, the insurance features, and the payouts are understood and
internalized. Our simulation analyses help illustrate how potential choices could have affected
financial outcomes had an investor purchased a given RILA option at a particular time in the
past. Ultimately, these simulations help examine the potential economic value and the risks
associated with RILA investment options. While some form of these simulations may be helpful
for investors to understand the implications of an investment option selection problem faced
when choosing among the specific insurance feature options studied, they may not reflect the
economic values and consequences over the course of the life of the contract. That is, these
types of simulations could be difficult for potential RILA investors to understand, because RILA
insurance features change frequently, making past performance of today’s features irrelevant to
current investors who are not able to access those particular features in future market conditions.
This is particularly true at the end of an investment option’s investment term while the investor is
under age 59½ or within the surrender period: at future points in which they need to select new
investment options, today’s insurance protections may have been replaced with a less favorable
set of features. Of course, our simulations are not conducted with “today’s” features, but rather
with the features offered as of our early 2023 data construction.
These analyses are important because there is very limited research on RILA products,
their performance, and the extent to which they serve investors’ needs. We are aware of only a
few academic research studies on RILAs. These studies focused on understanding why RILAs
may be popular with investors (Mezger, 2023) and how RILAs compare to fixed-income and
variable annuities (Moenig, 2021). 10 Neither study fully explored the simulated performance of
10
A recent paper also discusses topics related to RILAs, but the context of this work is not germane to the current research. See Ellis, Moenig, &
Wise (2023).
Investor Testing Report on Registered Index-Linked Annuities | 10
RILAs nor how investors’ understanding of RILAs may vary when reading various disclosures,
both of which are topics that we examine in this report. Although historical performance is no
guarantee of future investment results and the insurance features available in the future may
change, historical performance simulations allow us to document how certain product features
offered as of the time we constructed our data may affect investors’ outcomes based on the
historical record so that we may study the potential usefulness of the products to investors and
gain a better understanding of the risk–reward tradeoff that RILA options offer. That
understanding will remain somewhat incomplete so long as historical data on feature offerings is
unavailable to researchers.
Data
We used data from Morningstar Intelligence Annuity Tool, which were extracted in
March 2023 (Morningstar, 2023a). The data included all RILAs that were open for new
investments as per Morningstar’s records of that date; our list of offerings may differ somewhat
from other sources, such as the SEC’s EDGAR database, because sources can use their own
inclusion criteria. 11 The data contained information on the investment options, including
insurance features (buffers, caps, triggers, participation rates, and floors). Because, to our
knowledge, these data have not been used in prior academic research, the data were validated by
manually reviewing prospectus reports. To better understand the performance of the underlying
indexes, we merged this RILA information with index returns data, which we extracted from
Morningstar Direct (Morningstar, 2023b). Our index returns data covered the period from 1990
through 2019.
Methodology
To provide a manageable set of comparisons for simulating historical RILA performance,
we focused on a common index, investment term, and insurance feature combination: an S&P
500 price return–linked index with an 18% cap, 10% buffer, and a 1-year investment term. We
compared point-to-point returns under this investment option with the performance of the S&P
500 price return–linked index, which allowed us to highlight how insurance features may work
to reduce downside risk while limiting upside potential. Some RILA providers illustrate the
impacts of the insurance features with graphs that compare returns of the underlying index to the
returns one would receive with a RILA. An investment in an index (e.g., through an index fund)
would typically track a total return index, which would have greater returns than a price return–
linked index due to dividends. However, comparing RILA returns to price return–linked indexes
more clearly illustrates how RILA insurance features affect outcomes relative to the index
returns.
11
EDGAR contains information based on RILA registrations, but registration does not necessarily mean the RILA is in operation and accepting
money from investors.
Investor Testing Report on Registered Index-Linked Annuities | 11
Note that analyses in this section were chosen to illustrate how a RILA could work and
are not intended to be representative of actual investment performance during this period. In
particular, we concentrated on a common investment option offering and traced out the
hypothetical performance of these insurance features had they been offered in prior years.
To be explicit, our simulations provide a representation of how particular investment
options would have evolved in value during the historical period. The simulations do not suggest
a purchaser’s actual returns during this period, because the insurance features may have evolved
over time. For example, a person investing in a 1-year investment option with an 18% cap and
10% buffer may face a RILA investment menu with a different set of features after 1 year, in the
case that the original 18% cap option is discontinued. Historical simulations for that new
investment option could, therefore, show a markedly different historical simulation return profile
than the original option—one that could be more or less favorable for the investor. And, unlike
some other investment vehicles, the investor cannot simply walk away from the RILA contract
without incurring surrender charges or other penalties. Because data on such variation of the
insurance features over time are unavailable, even the savviest investor has limited tools
available to ascertain the value of the RILA contract over the life of the contract. In essence, our
simulations help understand the value of the momentarily available investment options, but they
do not help shed light on the overall economic value and risks of the RILA contract over the
investor’s entire investment period.
Furthermore, the analyses discussed in this section did not address costs due to
withdrawing money from an investment option during an investment term, surrender charges, or
any tax implications (due to tax deferral or tax penalties for liquidating before the holder is 59½
years old). The assumption was that the investor purchased the RILA with the associated
investment option on a given date and held it until the end of the investment term; the investor
may have had to put the resulting funds into another investment option at the end of the term to
avoid surrender charges or other penalties. Research on 401(k) retirement plans has suggested
that withdrawals from retirement plans before the age of 59½ may be fairly common, with as
many as 40% of retirement plan participants cashing out retirement savings when leaving an
employer (Armour et al., 2017; Lu et al., 2017; Wang et al., 2022). If withdrawals from RILAs
are similarly common, our results may overstate the returns from RILA investments because the
analyses did not account for fees, charges, and penalties.
Limitations
Caps and triggers vary frequently over time, and we are not aware of any data source that
compiles historical rate sheets. Without historical data on insurance features, we cannot say
whether these products would have been common—or even available—in the years since RILAs
began being offered. Additionally, we cannot compare available rates on insurance features to
historical rates, so the simulations may have over or understated RILA performance relative to
what would have been available for an investor at the time of purchase. RILA products were not
offered before 2010, so the performance simulations used periods during which RILAs were not
Investor Testing Report on Registered Index-Linked Annuities | 12
available in order to provide a more complete picture of how insurance features related to the
historical performance of the accompanying index. (After all, for example, an investor evaluating
a 10% buffer on the S&P 500 might reasonably be interested in how historically common it is for
the S&P 500 to lose more than 10%, and that historical question would reasonably not be
confined to the period in which a RILA product of that type was offered). 12 These limitations
complicate the process of validating our data and highlight a critical lack of information about
this market, potentially hampering investors’ ability to obtain and make use of decision-relevant
information for evaluating the economic value of and comparing products before they purchase
and manage their investments. To the extent that inaccuracies remain in the data, reported
statistics may have been affected; however, our individual simulations have reasonably
approximated the performance of the stated insurance features over the historical period.
Despite these caveats, these simulation results provide a useful perspective on historical
performance, allowing us to demonstrate how a particular (current) investment option would
have performed during certain historical periods; in principle, this type of information could be
informative for a potential investor.
Descriptive Statistics on Overall RILA Market
At the time of these data, 19 insurance companies offered RILAs, for a total of 71 RILA
products and 1,832 different investment options. The number of investment options available for
some contracts can be large, which may raise concerns about choice overload, especially because
investors are unlikely to have previous experience with RILAs (Scheibehenne et al., 2010). In
addition, when selecting from investment menus of this type, naive diversification (spreading
money equally across menu options) is often a strategy that investors pursue (Benartzi & Thaler,
2001; Scholl, 2022), and the negative implications of too many choices in an investment context
has been documented extensively (see review in Scholl & Fontes, 2022). Furthermore, most
insurance companies only began offering RILAs within the past decade (with market entry
between 2010 to 2023), meaning that there is limited performance history for these products.
RILA Indexes, Investment Terms, and Insurance Features
Each investment option comes with an underlying index. Figure 1 shows the frequency
of the underlying offered indexes. As shown, common indexes are the S&P 500, Russell 2000,
MSCI EAFE, and NASDAQ-100 price return indexes. These four indexes comprise over 80%
of the indexes underlying investment options available in the market.
Although this section presents the performance of investment options over realized historical periods, we also performed an analysis that
simulated random draws of the index’s return based on the historical distribution of index returns. The results presented in this section are similar
to those constructed based on random draws using historical returns.
12
Investor Testing Report on Registered Index-Linked Annuities | 13
Figure 1. Indexes Offered.
The investment term length specifies the period during which the investment returns are
calculated. Figure 2 shows that the most common term lengths are 1, 3, and 6 years. Investment
options with 2- and 5-year terms are also available, but these are relatively rare.
Figure 2. Term Lengths Offered.
Investor Testing Report on Registered Index-Linked Annuities | 14
Investment options are typically associated with insurance features that can limit gains
and losses, as described above. In Figure 3, these insurance features are categorized and
described in terms of upside and downside combinations. As shown, the combination of a cap
and a buffer is especially common, capturing nearly 80% of all investment options offered.
Figure 3. Insurance Feature Combinations Offered.
Figure 3 does not show the amount of insurance protection that is provided. In general,
to offer more protection (e.g., a larger buffer percentage or a smaller magnitude floor), insurance
companies usually impose a lower limit on gains. Tables 1 and 2 show the relationships between
cap levels and the corresponding buffers or floors. These tables are limited to investment options
with 1-year terms that are linked to the S&P 500 price return index, given that these investment
options are relatively prevalent.
Both tables show that, in general, higher cap rates come with smaller downside
protection. For example, cap rates between 10% and 20% are associated with buffers of up to
25%. In contrast, cap rates between 20% and 30% are only associated with buffers of 10%
(Table 1). Table 2 displays a similar relationship; for example, cap rates that are less than 10%
may limit losses entirely (i.e., a floor of 0%), but cap rates between 10% to 20% are most
commonly associated with a floor of -10%.
Investor Testing Report on Registered Index-Linked Annuities | 15
Table 1. Comparing Caps and Buffers for Investment Options With 1-Year Terms and the S&P
500 Index.
Caps
10% or less
10.1% to 20%
20.1% to 30%
Total
10%
1.5%
38.2%
20.6%
60.2%
15%
1.5%
17.6%
0%
19.1%
Buffer
20%
25%
0%
0%
10.3%
5.9%
0%
0%
10.3%
5.9%
30%
2.9%
0%
0%
2.9%
40%
1.5%
0%
0%
1.4%
Total
7.4%
72.0%
20.6%
100%
Table 2. Comparing Caps and Floors for Investment Options With 1-Year Terms and the S&P
500 Index.
Caps
10% or less
10.1% to 20%
20.1% to 30%
Total
-20%
0%
6.6%
3.3%
9.9%
Floor
-10%
0%
40%
0%
40%
-5%
0%
3.3%
30%
3.3% 13.3%
0%
0%
6.6%
43%
Total
33.3%
63.3%
3.3%
100%
Although there are tradeoffs between these features, Tables 1 and 2 also show there is
considerable variation in the caps offered for a given buffer or floor across providers. Some of
these differences could be attributed to differences in other dimensions of the RILA that are not
shown. For example, the Interim Value Adjustments could vary between RILA providers.
Finally, the types of insurance features vary by investment term. Figure 4 shows the most
common four combinations of insurance features that are available for options that have 1-, 3-, or
6-year terms. As shown, triggers are generally only available for investment options with 1-year
terms (although there is a single offering for a 6-year term). The cap-and-buffer combination is
the most common insurance feature for all three of these term lengths.
Investor Testing Report on Registered Index-Linked Annuities | 16
Figure 4. Insurance Features Offered by Term Length.
Illustrative Examples Comparing Investment Option Value to an Index
To illustrate how insurance feature choices can affect the contract value, Figure 5
presents the value of a $10,000 investment in our chosen investment option compared with an
investment (outside of a RILA) in the S&P 500 index from 1999 to 2005 (as one might have
realized through investment in an S&P 500 index fund). 13 Throughout this analysis, we used the
price returns of the S&P 500 (or other relevant indexes), not total returns. This means that the
returns on the index shown do not include any dividends that would have been received if one
invested in the index outside of a RILA. As a result, our analysis understated the potential
positive returns on an investment in an index. We note that the return earned from a RILA
typically is based on the price returns, not the total returns of the index, but it is at the discretion
of the insurance company.
From 1999 to 2000, the S&P 500 rose by 20%, but the RILA’s returns would have been
capped at 18%; thus, by 2000, the RILA would have slightly trailed the index. The situation
changed from 2000 to 2002, when the S&P 500 dropped in value. Over that period, the RILA’s
buffer would have reduced losses, allowing its cumulative value to exceed that of the
hypothetical S&P 500 investment. By 2005, at the end of the graph, the investment in this RILA
13
Note, however, that index fund fees are not considered.
Investor Testing Report on Registered Index-Linked Annuities | 17
investment option would have increased by 27% (ending at $12,730), whereas the index lost
1.5% (ending at $9,857).
Figure 5. Simulated Value of a $10,000 Investment in the RILA and Index From 1999 to 2005.
As a comparison, Figure 6 shows the same RILA held for a 6-year period starting on
January 1, 1991. In this instance, at the end of 6 years, the RILA would have underperformed
the S&P 500 index due to the 18% cap, which would have limited the returns in certain years
when the index gained value. The final investment value of the RILA in 1997 would have been
approximately $4,000 lower than the index. Together, these two figures illustrate the fact that
the value of a RILA (and the value of a RILA relative to its index) can vary significantly
depending on the time period during which it is chosen, even holding constant the insurance
features.
Investor Testing Report on Registered Index-Linked Annuities | 18
Figure 6. Simulated Value of a $10,000 Investment in the RILA and Index From 1991 to 1997.
RILA Performance During Various Market Conditions
To provide a more general view of how the relative performance of the RILA and index
(as shown in Figures 5 and 6) could have varied in different time periods, we explored a range of
hypothetical starting investment dates. Figure 7 presents a histogram of 6-year RILA and index
returns starting each month from 1990 to 2013. The pink bars capture returns for the index,
whereas the overlaid blue bars present simulated returns for the RILA. Specifically, the
histogram includes 276 return values, each of which represents a 6-year term starting in a
particular month: January 1990 to January 1996, February 1990 to February 1996, and so forth,
through January 2013 to January 2019. Although much of this time period includes periods
when RILAs were not available (since RILAs were first offered in 2010), this analysis illustrates
how returns can vary depending on the insurance features and the timing of the investment; they
also illustrate how the insurance features relate to the historical performance of the index. As
with Figures 5 and 6, the analytic assumption was that the 1-year RILA investment option was
reinvested in an identical investment option for six consecutive 1-year investment periods.
Investor Testing Report on Registered Index-Linked Annuities | 19
Figure 7. Simulated Ending Values of $10,000 Investments After 6 Years.
Figure 7 illustrates that the insurance features of the RILAs we have been considering
(i.e., the 18% cap and 10% buffer) have the potential to attenuate the magnitude of both losses
and gains. In particular, the height of each bar is proportional to the probability of that realized
investment outcome. It is apparent that the distribution of the RILA returns in Figure 7 is more
condensed than the distribution of index returns, with fewer observations with extremely positive
or negative values. This result suggests that this RILA investment option has a narrower range
of investment outcomes in the way one would expect: the purchaser experiences smaller gains,
but also fewer losses of a smaller magnitude.
Although Figure 7 presents the distribution of returns, it does not illustrate how the
market conditions related to the performance of the RILA investment option and the index.
Extending this analysis, Figure 8 allows us to examine the simulated performance of the
investment option (yellow line) against the corresponding performance of the index (black line);
the investment option’s relative gain or loss against the index’s performance is also provided (red
line). To be more precise, each point on the yellow line represents the final investment value
that would have been obtained by investment in the 1-year term, 18% cap, 10% buffer RILA
investment option on the specified date, if it were rolled over for 6 total years. For example, the
left-most 1990 value of approximately $19,332 shows that a $10,000 RILA investment with
these features made on February 1, 1990, would have achieved gains of $9,332 six years later.
The black line similarly represents the value of a 6-year investment in the corresponding index.
Thus, the performance presented in line graphs in Figures 5 and 6 are condensed in Figure 8 to
single points at the beginning of 1999 and 1991, respectively.
Investor Testing Report on Registered Index-Linked Annuities | 20
Figure 8. Simulated Historical Investment Values by Initial Investment Date.
Note. The figure compares the RILA investment option (1-year term, 18% cap, 10% buffer, invested for six successive 1-year
investment periods) and the index returns over the same 6-year investment period. The black line is the values of a $10,000
investment over a 6-year investment period in the index; the yellow line (triangles) is an analogous line for the RILA investment
option. The red line (circles) is the difference between the yellow and black lines. Dates correspond to the start date of the
investment. Results are not adjusted for inflation.
In Figure 8, we can see that the high returns for the 6-year periods starting in the early
1990s would have been limited by the RILA investment option. Conversely, the relatively poor
returns for the 6-year periods starting in the late 1990s would have been somewhat mitigated by
the buffer in our chosen RILA. For a hypothetical investor, that means they would have had a
very small chance of experiencing a negative nominal return from the RILA investment option
during these historical periods, but they would have also missed some of the strong returns that
the index experienced in the early and late parts of the historical period.
The difference between the index and the RILA investment option (red line) would have
been negative over periods of market growth (e.g., starting dates of 1990–1995) and greatest for
investments starting in 1994. Put another way, RILA investments would have had a good chance
of experiencing a marginally better performance than the index, primarily during the 6-year
periods starting in the late 1990s to late 2000s. The RILA investments also would have reduced
investors’ chances of a slightly negative return during that same period (as losses maxed out at
15% but average losses during loss periods were only 6.8%). However, the RILA investments
would have foregone considerable investment gains for the 6-year periods starting in the early
1990s and following the 2008 financial crisis. Figure A1 in Appendix A shows an analogous
graph for a 12-year investment period (i.e., rollovers for 12 years) and finds similar patterns with
respect to the RILA investment option and the index, insofar as returns and relative returns
would have varied over time.
Investor Testing Report on Registered Index-Linked Annuities | 21
Comparing RILA Features
Thus far, we have explored a RILA investment option with one set of parameters (1-year
investment term, 18% cap, 10% buffer, and linked to the S&P 500 price return index). However,
as noted above, there is tremendous variation in indexes, investment terms, and insurance
features available to potential investors that may affect their outcomes (Figures 1–4). In this
section, we explore how differences in these choices can affect simulated returns.
We identified a set of RILA investment options with different features than our
previously considered case. Our intention was to vary a single feature at a time to isolate the
effect of one change, such as a change in the level of a cap holding the buffer constant.
However, for many RILA options, multiple dimensions of the bundle of features change together
(as shown in Tables 1 and 2). Additionally, we attempted to select from a set of investment
options offered by companies that also offered our focal option (i.e., 1-year investment term, the
18% cap, 10% buffer, and linked to the S&P 500 price return index). Our intent was to capture
realistic alternative options that an investor may encounter. All investment options were linked
to the S&P 500 price return index, so that the differences would be due solely to their other
features. In the end, we identified the following alternative investment options:
1. An investment option with a 15% buffer (instead of a 10% buffer), thus providing more
protection on the downside. As a consequence of this greater protection, the cap for this
option decreased to 13.5%.
2. An investment option with a trigger instead of a cap. The trigger was 10.9%, with a
corresponding threshold of 0%. The buffer remained at 10%.
3. An investment option with a floor of -20% (instead of a buffer of 10%) and a cap of
19.5%. This option was chosen because it offered the closest cap to the RILA presented
in this section among all investment options with a floor.
In addition, we considered the role of the investment term. In general, caps tend to get larger
with longer investment terms, since cap rates are described in terms of overall changes over the
term (not annualized rates). Also, since market returns have historically often been positive in
the long run, returns for RILAs with longer investment terms are less often in the range for
which a 10% buffer would reduce losses. As a consequence, higher caps can be offered. To
examine the role of RILAs with longer investment terms, we also simulated:
4. An investment option with a 3-year term maintaining a 10% buffer and a cap of 130%.
5. An investment option with a 6-year term maintaining a 10% buffer and a cap of 500%.
Variations in Insurance Features and Simulated Returns
The histogram in Figure 9 was constructed analogously to Figure 7. Because all
simulated investment options were linked to the S&P 500, the past returns of the index (pink
bars) are the same for all panels in Figure 9. The effect of increasing the buffer from 10% to
15% and decreasing the cap from 18% to 13.5% is seen by comparing Panels A and B. As
shown, the increased buffer and lower cap (Panel B) are associated with more compressed
Investor Testing Report on Registered Index-Linked Annuities | 22
returns. In particular, the frequency of returns over 80% would have reduced from 35% to 18%,
and the frequency of returns over 20% would have increased from 79% to 85%. Both the 10%
buffer and 15% buffer examples (Panel A and Panel B) would have reduced the likelihood of
losing money to about 1% compared to a direct investment in the index (which, during the time
period from 1990 to 2013, lost money about 15% of the time; see Table 3 on page 29). On net,
this scenario presents a more conservative set of outcomes than our original investment option.
Figure 9. Distribution of Returns with Other Combinations of Insurance Features.
Investor Testing Report on Registered Index-Linked Annuities | 23
Panel C in Figure 9 provides the distribution of simulated historical returns for a RILA
investment option with a 10.9% trigger and a 10% buffer. This combination of features would
have provided the most downside protection out of our presented investment options; the
investment option featuring a trigger would have never lost value over a 6-year investment
period for any of our start dates. On the other hand, with a 10.9% trigger, the maximum return
over 6 years would have been 86%, which this option would have reached for 43% of the
simulation start dates. The index return exceeded the investment option’s return for 37% of the
start dates, which shows that it would not have been uncommon for the RILA option to miss out
on index gains.
Panel D provides the returns distribution for a RILA investment option with a 20% floor
and a slightly increased cap at 19.5%. This distribution is also relatively truncated, with the
maximum investment gain reaching $14,280. The distribution is bimodal, meaning that the
distribution shows two highly frequent outcomes: one in which investors would have lost money
relative to their initial investment, and one in which they would have gained money. Whereas
the probability of a loss with direct investment in the index was 15%, the probability of losing
money was 37% under this combination of features. This difference is because small losses
(which would have been relatively common) would have been incurred by the forced liquidation
of the investment at the end of the term (whereas with many investments, such as in an index
fund, the loss remains a paper loss until the investor decides to sell). The highest gain for the
option in Panel D was 43%, which it would have reached 38% of the time. Clearly, however, the
cap would have limited upside returns relative to the index.
Figure 10 traces the historical simulation return results over time for these three variants
on our baseline case. The line graphs display the same data as the histograms but offer a
different view in that they highlight the relationship between index returns and investment option
returns over time so that it is clearer how the investment option would have performed given a
particular realization of index performance. Panels A and B suggest a moderating of returns, but
the option presented in Panel C highlights an example with very little historical downside
protection, an increase in the probability of losing money, and a very limited upside in terms of
returns.
Investor Testing Report on Registered Index-Linked Annuities | 24
Figure 10. Simulated Historical Investment Values by Initial Investment Date, Comparing
Different Insurance Features.
Panel A
Note. Panel A compares the RILA investment option and the index returns over the same 6-year investment period. The black
line is the values of a $10,000 investment over a 6-year investment period in the index; the yellow line (triangles) is an analogous
line for the RILA investment option. The red line (circles) is the difference between the yellow and black lines. Dates correspond
to the start date of the investment. Results are not adjusted for inflation.
Panel B
Note. Panel B compares the RILA investment option and the index returns over the same 6-year investment period. The black
line is the values of a $10,000 investment over a 6-year investment period in the index; the yellow line (triangles) is an analogous
line for the RILA investment option. The red line (circles) is the difference between the yellow and black lines. Dates correspond
to the start date of the investment. Results are not adjusted for inflation.
Investor Testing Report on Registered Index-Linked Annuities | 25
Panel C
Note. Panel C compares the RILA investment option and the index returns over the same 6-year investment period. The black
line is the values of a $10,000 investment over a 6-year investment period in the index; the yellow line (triangles) is an analogous
line for the RILA investment option. The red line (circles) is the difference between the yellow and black lines. Dates correspond
to the start date of the investment. Results are not adjusted for inflation.
Variations in Term Length and Simulated Returns
Turning to the effect of varying the term length, Figure 11 presents returns under 1-, 3-,
and 6-year terms, assuming a total investment period of 6 years. Note that, as term lengths grow
longer, the caps also tend to increase to accommodate potentially higher growth over a longer
period of time; cap rates are described in terms of point-to-point changes and are not annualized
growth rates. In practice, the higher caps are rarely applied because they would require
substantial returns in the underlying index.
A perhaps surprising result was that there would have been a greater chance of losses
with these examples under the 3- and 6-year terms (Panels E and F) than under the 1-year term
(each diagram again presents an investment for a total of 6 years). At the same time, the 3- and
6-year investment options would have tracked the upper part of the returns distributions quite
closely to the index; these results differ from the RILA investment option with the 1-year term,
which simulated returns that were significantly more likely to forego higher returns.
When comparing the 3-year and 6-year investment term options, there would have been
differences at values between $10,000 and $15,000. In particular, the 6-year investment option
would have created a concentrated mode in the $10,000 bin and would have had an increased
probability of a negative return. For this example, the 3-year term may have allowed for more
protection than the 6-year term, since the same 10% buffer could have been applied twice during
the investment period.
Investor Testing Report on Registered Index-Linked Annuities | 26
Figure 11. Distribution of Returns With Different Term Lengths.
Figure 12 illustrates the simulated returns for these comparisons by start date. In this figure, we
see that the 3- and 6-year products (with caps of 130% and 500%, respectively) would have been
largely unconstrained from capturing the full gains of the index during growth periods. They
also would have provided protection against some of the losses that would have occurred on the
downside, with the chance of a negative return much lower for the 3-year product than for the 6year product.
Investor Testing Report on Registered Index-Linked Annuities | 27
Figure 12. Simulated Investment Values by Initial Investment Date, Comparing 3- and 6- Year
Term Products.
Panel A
Note. Panel A compares the RILA investment option and the index returns over the same 6-year investment period. The black
line is the values of a $10,000 investment over a 6-year investment period in the index; the yellow line (triangles) is an analogous
line for the RILA investment option. The red line (circles) is the difference between the yellow and black lines. Dates correspond
to the start date of the investment. Results are not adjusted for inflation.
Panel B
Note. Panel B compares the RILA investment option and the index returns over the same 6-year investment period. The black
line is the values of a $10,000 investment over a 6-year investment period in the index; the yellow line (triangles) is an analogous
line for the RILA investment option. The red line (circles) is the difference between the yellow and black lines. Dates correspond
to the start date of the investment. Results are not adjusted for inflation.
Investor Testing Report on Registered Index-Linked Annuities | 28
Summary Performance Statistics
Table 3 reinforces these points with summary statistics about the simulated record of
these investment options compared to the index. The table presents the mean returns of a
hypothetical $10,000 investment, the mean gains over the index, the percentage of the time the
RILA investment would have underperformed the index, and the percentage of time the RILA
investment returns would have been above certain thresholds (columns “> 0%” through
“> 80%”). Our baseline investment option (1-year term, 18% cap, 10% buffer) worked in the
expected way relative to the index: lowering the probability of (typically small) losses and
shifting the distribution of returns so that, for example, the probability of a gain greater than 40%
was higher with the RILA investment option; however, the likelihood of a gain in excess of 80%
was attenuated.
The bottom two rows of Table 3 show the 3-year and 6-year investment options. It seems
that the 6-year term product would have had a similar returns profile as the index, whereas the 3year product would have increased the probability of no loss (> 0%) and the probability of a
small positive gain (> 20%) compared to the index. On average, investors would have seen a
higher investment performance from direct investment in the index (outside of a RILA) than with
most of these examples, but the RILA investment options would have helped avert small losses
that occurred at particular times. The floor insurance feature largely had more problematic
performance in the historical period: relative to the index, it was frequently more likely to incur
losses and to fail to achieve high investment performance. On average, an investor choosing this
product would have lost 45% of their initial investment compared with a direct investment in the
index.
Table 3. Summary Performance Statistics of Selected RILA Investment Option Examples.
Features
Mean
SD
Mean
Gain
SD
Gain
< Index > 0% > 20% > 40% > 60% > 80%
S&P 500 Index
$16,494 6,791
-
-
-
85% 58%
51%
46%
40%
18% Cap, 10% Buffer, 1-yr term
$15,878 3,968
-$616
3,125
45%
99% 79%
55%
48%
35%
13.5% Cap, 15% Buffer, 1-yr term $15,100 2,707 -$1,395
10.9% Trigger, 10% buffer, 1-yr
term
$15,256 3,032 -$1,238
4,415
51%
99% 85%
57%
44%
18%
4,509
47%
100% 78%
57%
44%
41%
19.5% Cap, 20% Floor, 1-yr term
$11,930 2,135 -$4,565
5,058
86%
63% 48%
40%
0%
0%
130% Cap, 10% Buffer, 3-yr term
$17,129 6,305
712
19%
98% 71%
53%
48%
41%
$635
500% Cap, 10% Buffer, 6-yr term $16,588 6,693
$94
257
34%
85% 58% 51% 46% 40%
Note. The mean column is the average end-dollar amount of a $10,000 investment in the RILA for a 6-year rollover period across
all start months. The columns with “gain” compare the RILA returns to the index. The “< Index” column shows the percentage
of start months that the RILA would have underperformed the index. The “> 0” column shows the percentage of start months for
which the end-investment level would have been greater than the initial investment. The “> 20%” column shows the percentage
of start months for which the return would have been greater than 20%.
Investor Testing Report on Registered Index-Linked Annuities | 29
Once again, we stress that these examples are provided to illustrate that RILA investment
options can have nonintuitive implications, such as an investment option with a floor
underperforming an index in the majority of simulated historical cases. The examples were
chosen by identifying a common investment product option and examining deviations in features
from that option for the purpose of understanding how RILA products can serve investment and
investment–insurance functions. As illustrative examples, they are in no way intended to be
representative of how the insurance features work across a broader set of products.
Risk
The insurance protections that the RILA investment options provide suggest a natural
interest in RILA products’ ability to mitigate risk. The examples considered previously
suggested that some RILA investment options could be helpful at mitigating downside while
limiting the upper distribution of returns.
Standard economic and finance theory typically conceptualizes risk in terms of the
variation in outcomes such as the standard deviation of returns or the Sharpe Ratio. For some
investors, particularly retail investors, different formulations of risk may be relevant or more
intuitive (Holzmeister et al., 2020); for example, some investors might prioritize a definition of
risk related to the chance of losing investment dollars over a variation in returns, which in some
circumstances might represent variation in the degree of positive returns.
Bearing this in mind, in this section, we examine cases in which the RILA investment
option would have had a higher risk in terms of the variation in returns and in terms of an
increased probability of losing money. Of course, other variations of risk concepts might be of
interest to other potential investors, but we confined our discussion to the above for the present
analysis.
It is important to note that although in some cases, these higher risk levels for the
investment option may be adverse for the investor, in other cases, they may not be. For example,
Figure 13 provides an example of an investment option that would have demonstrated a much
higher implied volatility than the associated index. By visual inspection, for much of the
historical period, the RILA investment option would have outperformed the index, but there
would have been high variability in the realized returns across start dates. The investment option
would have avoided some of the index’s cumulative losses during the investment start dates in
the late 1990s and would have cushioned some of the index’s cumulative losses during the
investment start dates in the latter 2000s.
Investor Testing Report on Registered Index-Linked Annuities | 30
Figure 13. Examples of RILAs With Higher Risk Than the Underlying Index.
Note. The figure compares the RILA investment option and the index returns over the same 6-year investment period. The black
line is the values of a $10,000 investment over a 6-year investment period in the index; the yellow line (triangles) is an analogous
line for the RILA investment option. The red line (circles) is the difference between the yellow and black lines. Dates correspond
to the start date of the investment. Results are not adjusted for inflation.
Figure 14 provides an example of an investment option that, during the historical period,
would have increased the probability of loss while forgoing substantial gains. In addition, it
would have provided only slight protection against losses that would have been incurred for
investment terms originating in the late 1990s and very early 2000s. In fact, when experiencing
a loss, the average value of a loss for the investment option would have been about 3% higher
compared to the index. In other words, the investment option would have been more likely to
lose money and, on average, would have had a higher magnitude of loss when there was a loss
compared to the index (note that this is on average when the investment option or the index
independently dipped negative, so the “average loss if loss” magnitudes were not computed over
comparable time periods). This particular pattern would have emerged with, among other things,
the substantial gains in the NASDAQ followed by a large crash, which is displayed in Figure 14,
for the investment periods starting in the late 1990s. In these circumstances, the cap would have
greatly limited gains for periods starting in the 1990s, but the buffer would have only slightly
tempered losses experienced during the crash. Moreover, this example highlights some of the
cumulative effects of the insurance provisions over successive periods: Some initiation dates
would have resulted in a capping of the upside, followed by a realization of sometimes
successive years of downside. As such, foregone gains and losses would have compounded
considerably.
Investor Testing Report on Registered Index-Linked Annuities | 31
Figure 14. Examples of RILAs With Higher Risk Than the Underlying Index.
Note. The figure compares the RILA investment option and the index returns over the same 6-year investment period. The black
line is the values of a $10,000 investment over a 6-year investment period in the index; the yellow line (triangles) is an analogous
line for the RILA investment option. The red line (circles) is the difference between the yellow and black lines. Dates correspond
to the start date of the investment. Results are not adjusted for inflation.
Figures 13 and 14 make clear that some RILA investment options may increase
investors’ risk rather than mitigate investors’ risk compared to direct investment in the market,
even when early withdrawal penalties are not applied, which may be concerning for investors
given that RILAs are marketed as a risk mitigation device. Further research and more
transparency are needed in order to document the prevalence of such patterns in the marketplace
as a whole and determine the extent of any investor protection concerns they may raise.
Summary and Discussion
Overall, our results suggest that the choices that purchasers make about RILA investment
options can significantly impact returns. We also identified investment options that would have
demonstrated potentially adverse risk and return properties for investors, as reflected in
simulated historical performance.
These results underestimate the possibility that actual investors would have experienced
investment losses over the historical simulation period with RILAs to the extent that we did not
account for investor withdrawals before the end of an investment term (which may incur Interim
Value Adjustments), before the end of the surrender period (which may incur surrender charges),
or before age 59½ (which would incur tax penalties). Previous research has found that as many
as 40% of 401(k) retirement plan participants liquidate their accounts upon job transition
(Armour et al., 2017; Lu et al., 2017; and Wang et al., 2022), suggesting that using retirement
funds for other purposes can be common. If investors were to withdraw money from their RILA
Investor Testing Report on Registered Index-Linked Annuities | 32
contracts as often as they do with their retirement funds, mid-term withdrawal penalties and
surrender charges could apply with some frequency. Further research and data transparency are
needed to understand and monitor the impact of such penalties on investors. For example, if in
the extreme, most RILA investors need to liquidate a RILA investment, incurring associated tax
penalties, Interim Value Adjustments, and surrender charges, these charges and penalties could
limit the economic value of the products, or create other balance sheet stresses for households.
Given that some penalties may be significant, one would hope that they are rarely levied in
practice; even if the penalties paid do not affect the financial system systemically, their impact
on individual investors could still be significant. Data on RILA withdrawals and penalties are
thus likely to be relevant to investors, financial intermediaries, and other stakeholders. Such data
seem critical to an investor’s ability to make informed investment decisions, and for
understanding the extent to which RILA recommendations result in beneficial outcomes for
investors.
Investor Testing Report on Registered Index-Linked Annuities | 33
4. Review of RILA Marketing
To better understand RILA products and the information investors may receive, we
reviewed the websites of 13 insurance companies that sell RILAs. Although this was not a
review of all RILAs available in the market, it provided us with a picture of how RILAs are
marketed to consumers. These companies were identified using the 2021 Secure Retirement
Institute’s U.S. Individual Annuities Sales Survey (LIMRA, 2022). This review included the
web pages, product brochures, videos, prospectuses, fact sheets, and other accompanying
promotional material available to retail investors. The reviewed marketing materials presented
the balance between the opportunity for growth and the reduced risk of losses on the investment
relative to savings alone or more conservative investments as one of the main benefits of a
RILA. If comparisons were made to the returns on the underlying index, they were typically
based on an example or a figure that highlighted the loss protection, not the caps on gains. The
reviewed materials were typically marketed toward those who are nearing retirement, featuring
photos and vignettes that suggested that the product is relevant for individuals nearing
retirement, although some materials discussed end-of-life issues for investors dealing with
terminal illness. Prospectuses were readily available for most companies, although some were
absent or challenging to locate. Among the 12 companies that had prospectuses available, the
number of pages in the prospectuses ranged from 75 to 262, at an average of 174 pages. The
length and language of these documents could be daunting to an average investor.
As we saw in the simulations in Section 3 of this report, insurance features (including
caps, triggers, floors, and buffers) are important choices for investors. Many of the reviewed
companies emphasized that RILAs can be customized to fit the investor’s needs. However, to
make informed choices, potential purchasers need to understand the offered features and their
actual implications for investing over long periods of time. Although all companies described
these features, they differed in how information about features was presented, using hypothetical
scenarios or vignettes, definitions, interactive tools, or graphs to illustrate the different insurance
features. The placement of information also varied, with some displaying information
prominently and others only providing technical information in a prospectus or rate sheet. Other
key topics such as early withdrawal charges, investment terms, and index options were included
in many but not all materials.
Investor Testing Report on Registered Index-Linked Annuities | 34
5. Qualitative Testing
We conducted two rounds of in-depth interviews to assess potential RILA disclosure,
most notably a hypothetical KIT, for areas of confusion or misunderstanding, and to gain insight
on investors’ mental models regarding the way RILA products function, including potential
benefits, risks, and drawbacks. Our qualitative testing used a semi-structured, open-ended
format so that participants could express their reactions and beliefs, regardless of accuracy. This
method is important for uncovering how everyday investors reason about RILA products and for
understanding their reactions to potential RILA disclosures.
Methods
To recruit participants for the in-depth interviews, the research team developed a
screening survey to target two groups of participants: annuity owners and annuity “shoppers.”
Owners reported currently having an annuity, whereas “shoppers” reported actively shopping for
—or planning to shop for—“a financial product to manage income flow after retirement.” 14
Additionally, participants needed to have at least $25,000 in retirement assets. We aimed for an
even mix of participants across demographic characteristics, locations, and (in Round 2 only)
mutual fund literacy (Scholl & Fontes, 2022). 15 Potential participants expressed interest in
participating in a 1-hour virtual interview on annuities. In total, we recruited nine participants
for the first round of interviews and 11 participants for the second round of interviews. In
Round 2, several participants were significantly more sophisticated than the average investor,
having worked in the field of finance, potentially as investment professionals, and several
participants were high net-worth individuals. Participants consented to participate in the
interviews.
In both rounds, over the course of an hour, the interview proceeded with questions in five
major sections: First, there was a warm-up section during which participants were asked about
their experience with investments, which allowed moderators to ask general questions and
establish rapport. Second, participants were asked about their experiences with annuities and
RILAs, if any. All participants viewed a short introduction to RILAs. Third, participants were
shown select rows of a hypothetical KIT, one at a time. They were asked to read the displayed
text, identify areas of confusion, and discuss how that information affected their interest in
RILAs. The rows and content of the KITs differed in the two rounds of interviews, to cover
more topics across both rounds. Fourth, participants were asked to apply their knowledge of
RILAs by assessing the appropriateness of RILAs for hypothetical investors (both rounds),
Because of the relative newness of RILA products, their relatively narrow niche, the timeframe available for research, and the potential for
misidentification during the recruitment process, the research team did not deem it feasible or favorable to focus on current RILA owners. To
maintain focus on disclosure development, our intention was to focus on potential purchasers rather than others who might have substantial preexisting knowledge of the products.
15
Respondents had the following demographic characteristics: 11 were male and 9 female, the average age was 50.9, 11 were white, 4 were
Hispanic, 3 were Black, 1 was Native American, and 1 was Asian. 12 were currently annuity owners, and 8 were currently or planning to shop
for annuities. 5 of 20 either reported over $1 million in retirement assets or stated that they were very knowledgeable about personal finances.
14
Investor Testing Report on Registered Index-Linked Annuities | 35
definitions of key terms (Round 2 only), and tradeoffs between investing in a RILA and in a
stock market index (Round 2 only). The final section allowed for any final comments and
reminded participants that moderators had presented only limited information about RILAs, so
more investigation would be necessary before making an investment decision regarding those
products. Our interview guides were reviewed and approved by an Institutional Review Board
(IRB) that reviews research to ensure that research plans are ethical and follow applicable
policies.
During each round, pilot interviews were conducted to ensure that the protocol
functioned well and to train the moderators on the interview guide. All moderators had
interviewing and moderating training through either the RIVA Training Institute or the
University of Maryland Joint Program in Survey Methodology (JPSM). Round 1 interviews took
place between March 27th, 2023 and March 31st, 2023, with Round 2 interviews taking place
from June 5th, 2023 to June 16th, 2023.
Limitations
Note that our qualitative research had several limitations. Due to time limitations, we
asked participants to focus on specific information about RILAs. Thus, we were not able to
ascertain if participants would have been able to identify and apply key information from a
longer KIT document that might include many pages. In Round 1, participants received an
informative introduction to RILAs, which may have increased their RILA understanding
compared to a typical investor. However, in Round 2, participants received different information
about RILAs to keep our focus on participants’ interpretation of the material directly presented in
the KIT. In both rounds, the interview participants were neither nationally representative nor
randomly selected; thus, the findings should not be considered representative of all consumers or
used for statistical analysis.
Results From Round 1
In Round 1, participants discussed four rows from a hypothetical KIT: “What is the
purpose of the contract?” “Who may the contract be appropriate for?” “Are there charges for
early withdrawal from the contract?” and “Are there other penalties for mid-term withdrawals?”
(See Table B1 in Appendix B.) For the purposes of this report, we highlight three broad
categories of results from these interviews: (1) whether participants could identify when a RILA
product may be appropriate, (2) difficulties with key RILA terminology, and (3) perceived
limited information.
We used scenarios to assess whether participants could determine the appropriateness of
a RILA product for hypothetical investors (see Appendix B for scenario text). The scenarios
attempted to emphasize a mix of RILA product features, including their potential benefits,
potential charges and penalties for hypothetical purchasers, and other factors. In many, but not
all cases, participants were able to identify the appropriateness of a RILA for investors in the
Investor Testing Report on Registered Index-Linked Annuities | 36
scenarios, although some scenarios were easier than others. In their explanations, participants
often identified only a single charge or penalty that would apply (i.e., only surrender charge,
early withdrawal tax penalty, or Interim Value Adjustment), rather than all the losses that would
occur (e.g., one scenario involved all three charges). This finding raises the concern that
participants demonstrated only a partial understanding of the relevant RILA penalties and how to
quantify them. However, for the most part, participants recognized that there would be penalties
associated with early withdrawal and that a RILA product would not be appropriate for people
with a short-term investment horizon or the potential need for liquidity.
Participants struggled to explain and understand key jargon used to describe RILAs.
Many reported that specific terminology is confusing, including language such as “index,”
“Interim Value Adjustment,” and “buffer.” In particular, the concept of the investment term
seemed to create significant and highly consequential confusion for participants. Several
participants seemed to equate “investment term” or “term” with the length of the insurance
contract rather than the length of the investment product options within the RILA contract; in
other words, they thought that they could liquidate their investment and get cash back after only
1 year without incurring any penalties. When discussing the Interim Value Adjustment and
buffer, participants generally reported that they did not understand these concepts. Overall, this
difficulty with RILA-related jargon is consistent with past research showing that investment
jargon can engender significant confusion in participants (Chin et al., 2023).
Several participants said they would need to do more research or would need more
specific information to make a decision about investing in a RILA. Areas of interest included
more specific information about the index being used; the amount of the cap, floor, or buffer; or
a graph or chart that might help them understand the concept of a floor and buffer.
These Round 1 results informed several modifications to the hypothetical KIT and
follow-up research questions for quantitative testing. In particular, before Round 2, the KIT was
modified to use the phrase “investment term” rather than “term,” to include a table to show how
the investment term interacts with the contract length, to examine various graphs to provide more
information about RILA insurance features like buffers, and to include expanded links to
additional information to indicate that more information could be available. As we discuss
below, we chose to include questions about key areas of confusion, such as the long-term nature
of RILAs and RILA jargon, in quantitative testing.
Results From Round 2
In Round 2, participants discussed five rows from the KIT: “What is the purpose of the
contract?” “What is the time period for measuring growth (or loss) on my contract value?” “Are
there charges for early withdrawal from the contract?” “Are there other penalties for mid-term
withdrawals?” and “Is there a risk of loss from poor market performance?” (See Appendix C.)
Three of these topics overlapped with those covered in Round 1, but the presentation and content
Investor Testing Report on Registered Index-Linked Annuities | 37
of the information differed based on modifications to the KIT made between the rounds. The
results from Round 2 of the qualitative testing fall into five broad categories: (1) concern about
penalties stemming from mid-term withdrawals, (2) understanding the long-run nature of RILAs,
(3) difficulties with key RILA terminology, (4) perceived limited information, and (5) difficulties
with introductory information about RILAs.
Regarding charges, the KIT that was tested in Round 2 emphasized that significant
charges can be incurred if money is withdrawn mid-term. Nearly all participants were concerned
about the “up to 90%” Interim Value Adjustment that was displayed in the KIT (row titled
“Interim Value Adjustment maximum loss potential (% of contract value)” (see Table C1 in
Appendix C). Participants indicated wanting to understand this text better and wanting more
information about how this value is calculated, as many expressed incredulity and could not
comprehend how an investment could have such a large loss or large charges associated with it.
However, attention to the amount of the charge was not an indicator of understanding. At the
time it was initially introduced, participants indicated being confused, and later in the interview,
they were generally not able to describe the Interim Value Adjustment.
Attention to the Interim Value Adjustment affected interview participants’ understanding
of the long-term nature of RILAs. In particular, after carefully reviewing the text about Interim
Value Adjustments, the 90% potential loss was so salient that most participants were able to
identify that a RILA is not a short-term investment and should only be used if an investor will
not make early withdrawals. This observation was particularly evident in the discussion of the
scenarios.
At the same time, participants indicated that they did not understand other contract
details. As mentioned above, several participants in Round 2 were significantly more
sophisticated than the average investor, having worked in the field of finance, and several were
high net-worth individuals. Even these participants struggled to correctly apply the concepts
discussed in the KIT.
With the goal of helping potential investors better understand key concepts that were
difficult for Round 1 participants, a table was introduced to illustrate how charges are applied
over the surrender period of the contract, the difference between the investment term and the
contract length, and how charges could vary over the course of the contract. This table was
confusing for even the most sophisticated investors, as participants could not understand the
potential total losses from charges and penalties nor distinguish between surrender charges and
the Interim Value Adjustment. 16 If participants were able to define the Interim Value
Adjustment when they initially saw they table, they were not necessarily able to define the
concept later in the interview or to apply their knowledge. Round 2 also introduced several
16
It is possible that alternative visualizations could yield different results; in these interviews, we only tested a single design.
Investor Testing Report on Registered Index-Linked Annuities | 38
graphics to illustrate how a buffer works. Most participants were unable to correctly apply the
definition of the buffer.
Participants indicated wanting information that was not part of the tested KIT rows, such
as the specific investment index options. This desire may have been an artifact of the more
limited information that was provided to participants about RILAs before viewing the KIT in
Round 2 compared to in Round 1. Some of the more sophisticated participants said they wanted
information about past RILA returns. Several participants wanted additional information about
other fees that might be charged. In this round, we did not focus on caps on gains, which may
have contributed to some participants wondering what other fees there might be, with some
participants noting that the insurance company has to make money somehow. Some participants
noted that if they needed additional information they would turn to a financial professional,
which speaks to the importance of understanding the incentives associated with sales of RILAs
and any potential conflicts of interest.
To introduce participants to the RILAs, all interviews contained a short explanation of
RILA contracts (see Appendix C). The introduction to the RILA contract that was provided in
the interviews did not always help people understand the purpose of a RILA or who a RILA
might be appropriate for. In particular, many participants did not seem to understand that RILAs
are intended as a retirement savings vehicle and that there may be tax penalties for withdrawal
before age 59½. To further explore how introductory information may help individuals
understand RILAs, we return to this question in the quantitative testing.
Summary of Qualitative Testing
Qualitative testing was instrumental to the research team’s ability to understand
participants’ responses to the RILA products and the provision of information about the
products. Across both rounds of interviews, we found that many interview participants struggled
to understand the details of the RILA contract presented in the hypothetical KIT. These
difficulties manifested both in their explanations of what they read and in how they applied the
information to the hypothetical scenarios. Often, participants reported that the information was
clear, but when they were asked to explain or apply that information later in the interviews, their
responses suggested incorrect interpretations. Specific terms such as “index,” “buffer,” and
“Interim Value Adjustment” were often cited as confusing, suggesting that RILA jargon causes
difficulties but also that the underlying concepts also create significant challenges. Furthermore,
many participants did not seem to understand the distinction between the investment term and
the contract length, which seemed to create additional downstream confusion: Participants
appeared to carry forward their mistaken understanding of contract length to their interpretations
of how insurance provisions or other product features work. However, after being asked to
concentrate on Interim Value Adjustments, they were able to identify that a RILA is not an
appropriate option for someone with a short-term investment horizon or for someone who may
need more liquidity. Across both rounds, many participants noted that they would need more
Investor Testing Report on Registered Index-Linked Annuities | 39
information to fully understand the product. The results of the qualitative interviews helped
inform the design of the quantitative study.
Investor Testing Report on Registered Index-Linked Annuities | 40
6. Quantitative Testing
Our quantitative testing was designed to assess whether the organization of the
hypothetical KIT and other information provided to participants impacts their comprehension of
the disclosed information. In particular, we sought to answer the following research questions:
(1) Does providing introductory text to a disclosure impact comprehension of the disclosed
information? and (2) Does the structure of the topic headings, either in a question-and-answer
(Q&A) format or statement format impact comprehension?
Interest in these research areas arose because of the overall complexity of the products,
the commensurate complexity of the KIT, and the findings of the foregoing analysis—
particularly our qualitative inquiry. Among other things, the qualitative research suggested
substantial difficulty in understanding basic product features, jargon, more technically
challenging features (e.g., definitions of insurance features and charges), and the implications of
such features in terms of risk, payouts, and other factors. The introduction variants we tested
were informed by the qualitative interviews, during which we found that the KIT did not always
provide participants with sufficient background information to understand RILAs. The Q&A
versus statement format was intended to test whether framing information as a question would
help participants identify and locate relevant information. Because of the difficulty of the
material, our design also included various alternative incentives to test whether incentivizing
people to read more carefully could improve comprehension. 17
Methods
Sample
We had 3,078 participants start the survey from the Prodege opt-in panel, and 2,564
completed it. 18 An additional 42 participants were excluded because they had the same IP
address as other participants. 19 The quantitative testing was fielded from June 16 to June 23,
2023. The average response time was 14.7 minutes. Participants were compensated for their
time, with compensation varying from $3 to $12, depending on whether the respondent was in a
particularly hard-to-reach group based on annuity ownership and financial experience, as well as
incentives provided based on the number of questions they answered correctly. The survey
targeted participants who were more likely to have some experience with financial products.
Participants consented to participate in the survey panel. Our experimental survey design was
reviewed and approved by an Institutional Review Board (IRB).
Table 4 summarizes the demographic characteristics of our sample, and Table 5
summarizes the sample’s financial characteristics. Compared to the overall U.S. population, our
Although such incentives are unlikely to be contemplated as a policy instrument, they enable us to gain some insight on the burden associated
with reading complex RILA disclosures. Furthermore, these incentives can help assess whether low levels of comprehension measured in a
survey occur because the material is too difficult or because participants are not motivated to attend to the details.
18
For more information on the Prodege panel, see https://www.prodege.com/.
19
Non-independent observations, such as those arising from a single participant completing the study twice, violates one of the assumptions of
ordinary least squares regression analysis used below.
17
Investor Testing Report on Registered Index-Linked Annuities | 41
sample was more highly educated, more likely to be female, older, and more likely to be White.
Given the potential difficulty in recruiting annuity owners—a particular group of interest—we
did not formally target demographic representativeness in this study. We asked additional
financial background questions to measure the financial characteristics shown in Table 5.
Approximately 68.4% of our sample had income under $100,000 per year, 25.5% had income
between $100,000 and $200,000, and 6.1% had income greater than $200,000. Approximately
7% of our sample reported being in debt, 12% reported being neither in debt nor having any
assets, 22% reported having positive but less than $100,000 in assets, 43% reported having assets
between $100,000 and $1,000,000, and 16% reported having assets greater than $1,000,000. We
found that 62% of our sample were classified as an Independent Investor, meaning that they held
some financial assets outside of a retirement account; 31% of our sample were classified as a
Retirement-Only Investor, meaning that they held financial assets only in a retirement account;
and approximately 7% were Non-Investors.
Table 4. Summary of Demographic Sample Characteristics.
Overall
Education (%)
BA or above
Some college
High school or equivalent
No high school diploma
Missing
Gender (%)
Female
Male
Nonbinary
Missing
Age (mean [SD])
Race and Ethnicity (%)
American Indian or Alaska Native
Asian
Black or African American
Hispanic
Pacific Islander
White
Some other race
Prefer not to answer
Missing
Count (%)
2,522
1,710 (68.1%)
456 (18.2%)
308 (12.3%)
36 (1.4%)
12 (0.5%)
1,492 (62.2%)
900 (37.5%)
7 (0.3%)
123 (4.9%)
57.79 (13.83)
11 (0.5%)
117 (4.9%)
155 (6.5%)
143 (6.0%)
2 (0.1%)
1,882 (78.5%)
26 (1.1%)
60 (2.5%)
126 (5.0%)
Investor Testing Report on Registered Index-Linked Annuities | 42
Table 5. Summary of Sample Financial Characteristics.
Household Income (%)
<$5,000
$5K –$14,999
$15K–$24,999
$25K–$39,999
$40K–$59,999
$60K–$74,999
$75K–$99,999
$100K–$149,999
$150K–$174,999
$175K–$199,999
> = $200K
Net Worth (%)
-$75,001 or more debt
-$50,001, to -$75K
-$40,001 to -$50K
-$30,001 to -$40K
-$20,001 to -$30K
-$10,001 to -$20K
-$5,001 to -$10K
-$1 to -$5K
0
$1 to $10K
$10,001 to $50K
$50,001 to $100K
$100,001 to $200K
$200,001 to $400K
$400,001 to $650K
$650,001 to $1M
$1,000,001 to $2.5M
> $2.5M
Missing
Investor Type (%)
Independent Investor
Retirement-Only Investor
Non-Investor
Owns Annuity (%)
Yes
No
Missing
Count (%)
28 (1.1%)
34 (1.3%)
124 (4.9%)
314 (12.5%)
417 (16.5%)
354 (14.0%)
453 (18.0%)
459 (18.2%)
108 (4.3%)
77 (3.1%)
154 (6.1%)
12 (0.6%)
11 (0.5%)
8 (0.4%)
6 (0.3%)
13 (0.6%)
38 (1.8%)
37 (1.8%)
22 (1.1%)
256 (12.3%)
103 (4.9%)
182 (8.7%)
168 (8.1%)
194 (9.3%)
239 (11.5%)
218 (10.5%)
248 (11.9%)
226 (10.9%)
100 (4.8%)
441 (17.5%)
1,564 (62.0%)
783 (31.0%)
175 (6.9%)
1063 (42.1%)
1,458 (57.8%)
1 (0.0%)
Note. The count of participants in each category is followed by the percentage of participants by category in parentheses.
Historical Prodege screening variables were used to target an oversample of annuity owners. Annuity ownership and investor
type classifications were constructed using survey measures within our study.
Study Design and Overview
To examine our research questions, we employed a 4x2x2 between-subjects experimental
design. We included four different introductions to RILAs, two formats for the KIT, and two
variants of variable compensation. Our four introductions were designed to mimic different
Investor Testing Report on Registered Index-Linked Annuities | 43
information that individuals might be presented with about RILAs before reading the KIT. The
first introduction was a control scenario in which participants were given no information about
RILAs, which provided a baseline comparator for the other introductions (“control”). The
second introduction was designed to introduce participants to key terminology and was similar to
the introduction used during Round 2 of the qualitative interviews (“key terms”). The third
introduction described potential benefits of owning a RILA, was informed by our review of the
marketing materials provided to consumers, and used similar wording to the wording that a
consumer might encounter while shopping for a RILA (“benefits only”). The fourth introduction
was designed to present participants with information that could influence the decision to
purchase a RILA, focusing on conditions that would help a consumer identify whether a RILA is
appropriate for them (“decision focus”). The decision focus introduction was intended to
provide a jargon-free introduction informed by the qualitative interviews, during which we found
that participants were hampered by jargon and often did not seem to understand basic features of
RILAs such as their use as a retirement investment, or other aspects relevant to when RILAs may
or may not be an appropriate investment.
After seeing the introduction, participants were presented with one of two sets of KIT
rows, with each row presented one at a time. In the first set, each row label of the KIT was
framed as a question, whereas in the second set, the information was presented with each row
labeled as a statement. The material in the KITs was identical other than the Q&A or statement
format (see Appendix D for materials).
Next, participants answered a series of questions to gauge their understanding of the
presented material and its implications, and their ability to situationally apply the material in
settings that approximated key aspects of RILA decision making. Participants were also
randomly assigned to various incentive conditions. The study concluded by asking participants
for background information about themselves, such as demographics and financial asset
ownership (including annuities). Throughout the survey, we measured the amount of time spent
on each page in the background.
Comprehension Measures
Our main outcome of interest in this study was the comprehension of key concepts about
RILAs. Academic literature suggests that comprehension can be classified into multiple skills,
including basic comprehension and more advanced application of that knowledge (Bloom et al.,
1956); existing financial disclosure and financial literacy research also makes distinctions
between types of knowledge (e.g., Chin & Bruine de Bruin, 2019; Chin et al., 2022b; Scholl &
Fontes, 2022; Warmath & Zimmerman, 2019). As such, we made sure to include both factual
statements about RILA features and scenario-based questions that required participants to make
inferences about how RILAs would work for investors in certain situations. We included
multiple choice, true–false, and check-all-that-apply questions. The scenarios used in the testing
were based on those shown to participants in our qualitative interviews.
Investor Testing Report on Registered Index-Linked Annuities | 44
Our primary overall comprehension score is the percentage of comprehension questions
answered correctly. For the true–false questions, participants earned one point for correctly
selecting true or false. Participants could also select “Don’t know,” which was always scored as
incorrect. The check-all-that-apply questions were split into their individual response options
and scored one or zero for each possible response, due to the difficulty of these questions. For
example, a choose-all-that-apply question asking whether Interim Value Adjustments, surrender
charges, rollover fees, tax penalties, or none of the above applied in a certain scenario would be
recoded into five separate items. Participants could earn one point for each item by correctly
marking whether the charge, penalty, or fee applied. Incorrect responses, either for selecting the
wrong response or leaving the answer blank, would result in zero points for that item. This
recoding resulted in a total of 44 points that a participant could earn across all the comprehension
questions. Total points were then converted into a percentage of total points that a participant
earned. Given this recoding, someone guessing randomly among the response options would
expect to earn 22 points, or 50% of the total, on our overall comprehension measure.
Additionally, we created different sub-scores to examine four conceptual areas: (1) the
appropriateness of RILAs for investors based on their characteristics; (2) How a RILA Works;
(3) how the charges and penalties associated with RILAs affect liquidity; and (4) the insurance
protection offered by RILAs.
Moderating Characteristics
We considered several moderating variables: subjective financial literacy, investor status,
and ownership of an annuity. The subjective financial literacy scale was composed of three selfreported items measured on a 5-point scale: ability to deal with day-to-day financial matters,
assessment of math ability, and proficiency in dealing with saving and investing financial
matters. All three questions had outcomes from Strongly disagree to Strongly agree, with a
Don’t know option. Investor status questions asked participants about their ownership of
employer-sponsored retirement plans, then outside-employer retirement accounts, and finally
investments outside of any retirement accounts, such as a non-retirement brokerage or advisory
account (following the method discussed in Chin et al., 2022a; Chin et al., 2023). Participants
were directly asked if they owned an annuity, with some details of annuity products provided to
make the question easier to answer. If participants stated that they own an annuity, they were
asked which type of annuity (fixed, immediate, deferred, RILA, or I don’t know).
Limitations
There were several limitations to our study and analysis. First, we used an opt-in panel,
which did not allow us to make generalized statements about the comprehension of the
population as a whole. We prioritized distinguishing between annuity owners and non-annuity
owners and needed to use a panel large enough to target a significant number of annuity owners.
Second, we were limited in the disclosure documents that we could test based on both policy
priorities and our sample size. Alternative variations to the presented disclosures could have
impacted comprehension differently. Third, multiple rounds of quantitative testing were
Investor Testing Report on Registered Index-Linked Annuities | 45
infeasible because of time limitations and the need to incorporate qualitative testing before
designing the quantitative instrument. Best practices for research dictate starting with qualitative
testing to discuss RILAs with participants and to understand initial reactions to the disclosures.
Fourth, for practical reasons, we were unable to study the impact of a full KIT, which would
include many more rows and pages. Our study instrument needed to target an average response
time of 15 minutes for data-quality reasons. A KIT mock-up that included full information could
have taken the full 15 minutes (or more) to review in detail, leaving no time available for the
measurement of comprehension and application of knowledge. Fifth, as a new disclosure
proposal, the KIT that was used was not a KIT for a real RILA product. Product providers
generally have discretion in what information is presented, and as such, insurance companies
might choose to design KITs that differ substantially from what we presented. Sixth, our main
interest was to understand the information presented rather than all possible information about
investing and RILAs that may be important. Although comprehension is crucial for decision
making, we did not measure decision-making quality related to RILAs (neither regarding the
choice to purchase a RILA nor which insurance features to select). Finally, future product
innovations may change which features of a RILA have the largest impact on investor outcomes,
making additional and ongoing research important for knowing how disclosures convey the most
decision-relevant information.
Results
Across all questions and participants, the average percentage of questions scored as
answered correctly was 58% (see Table 6 for average sub-scores), which is statistically
significantly higher than 50% (one sample proportion test, p < 0.001), the score that would occur
through random guessing on true/false questions. To be considered fully informed, however, a
purchaser should arguably be able to answer a substantially higher percentage of the
comprehension questions. Furthermore, as shown in Tables 7–10, there was significant variation
in correct responses by question. For example, only 12.6% of participants correctly identified in
the How a RILA Works sub-score that RILAs are investment vehicles that are intended purely as
retirement savings vehicles rather than for achieving other goals (e.g., education and home
purchasing). Only 21% of participants correctly identified that a floor applies to a RILA
investment option rather than to the RILA contract overall (see Table 9). In contrast, 80.7% of
participants correctly identified that if an investor purchases a RILA, they cannot access their
money whenever they need it at no cost (see Table 10).
Table 6. Comprehension Overall and by Sub-score.
Overall comprehension
Appropriateness sub-score
How RILA Works sub-score
Insurance sub-score
Liquidity sub-score
Mean (SD)
58% (13)
57% (28)
49% (20)
57% (16)
62% (14)
Investor Testing Report on Registered Index-Linked Annuities | 46
Testing Impacts
We estimated the impacts of the different experimental factors on our main outcome
variables: overall comprehension and four separate conceptual sub-scores. 20 Full question text,
categories, and jargon classification are presented in Tables 7–10. To increase interpretability,
results are reported for analyses that did not include interaction terms between all experimental
factors but controlled for all of them using ordinary least squares (OLS) regression. The results
focus on both the relative differences between different experimental factors and the absolute
levels of comprehension, since comprehension is crucial for informed decision making. For most
of the analyses and unless otherwise specified, the control group was the group assigned to
receive no introduction (i.e., control introduction), the Non-Q&A KIT format, and no variable
incentives.
Table 7. Appropriateness Sub-score.
Question Text
Pat and Jean are in their early 40s. Their daughter is in seventh grade and will go to college
in six years. They have saved some money for college but are worried it won’t be enough,
so they would like to invest and make it grow. At the same time, they are not confident
about the stock market and are worried about losses.
Taylor is 50 years old with a steady job. She has a retirement account with stocks and
bonds that she started 30 years ago. Taylor doesn’t plan to use any of that money before
retirement. However, she is still worried about having enough given the rising cost of living
and the chance that the stock market could face a major drop in five or six years
Uncle Bob is 85 years old. He lives on his own, but expects to move to an assisted living
community in five years because he is increasingly forgetful. Assisted living is expensive,
and Bob is not sure if he’ll have enough money. He currently has some money in a
retirement account in very low-risk investments. His financial professional recommended a
RILA to him with a 5-year investment term. Uncle Bob thinks this will allow for some
stock market gains, with limited downside risk.
John is 55 years old and does not think he will ever have enough money to retire. He
doesn’t want to risk losses on his investments because if an emergency comes up, he will
need to withdraw that money and use it immediately. Right now he has a small,
conservative portfolio with mostly bonds.
Jargon
Average
Score
No
48.7%
No
54.1%
Yes
57.3%
No
69.1%
Note. Question text includes the type of question: True–false questions are labelled (T/F), multiple choice questions are labelled
(M/C), and choose-all-that-apply questions are labeled (CATA), with the individual option listed after the colon (e.g., CATA: Cap
Rate). Jargon indicates whether the question used jargon specific to RILAs. The average score is the percentage of all
participants who got the question correct.
There were additional preregistered analyses, which occasionally had different levels of significance, but the magnitude of the differences
between the testing impacts was quite small. The largest difference in testing effects between the outcome variables analyzed here and the
preregistered analyses was about 2 percentage points, which is small enough not to change the overall substantive conclusions.
20
Investor Testing Report on Registered Index-Linked Annuities | 47
Table 8. How a RILA Works Sub-score.
Jargon
Average
Score
No
12.6%
Yes
16.1%
Yes
32.9%
No
46.5%
Yes
51.5%
No
58.6%
A RILA does not have the same limits on contributions as a 401(k) or IRA. (T/F)
No
59.1%
RILAs cannot lose money. (T/F)
No
82.6%
RILAs are a long-term investment. (T/F)
No
83.9%
Question Text
RILAs provide an opportunity to save for a variety of investing goals (for example
education, homebuying, or retirement). (T/F)
The investment term is the life of your contract with the insurance company. (T/F)
At the end of the initial RILA investment term, you will receive a check from the insurance
for the current value of your investment. (T/F)
RILAs are among the safest annuity products one can buy. (T/F)
Your investment options are guaranteed to stay the same even if you stay in a RILA for
multiple terms. (T/F)
RILAs provide tax advantages for retirement savers. (T/F)
Note. Question text includes the type of question: True–false questions are labelled (T/F), multiple choice questions are labelled
(M/C), and choose-all-that-apply questions are labeled (CATA), with the individual option listed after the colon (e.g., CATA: Cap
Rate). Jargon indicates whether the question used jargon specific to RILAs. The average score is the percentage of all
participants who got the question correct.
Table 9. Insurance Sub-score.
Jargon
Average
Score
Yes
13.4%
Yes
16.2%
Yes
20.9%
Yes
51.6%
What features in RILAs reduce potential gains from the market? (CATA: Cap Rate)
Yes
52.1%
What features in RILAs reduce potential losses from market declines? (CATA: Buffer)
A RILA may limit your investment gains compared to investing in a fund that tracks the
same market index. (T/F)
What features in RILAs reduce potential gains from the market? (CATA: Buffer)
Yes
54.4%
No
62.6%
Yes
76.8%
What features in RILAs reduce potential gains from the market? (CATA: Protection Level)
Yes
81.3%
What features in RILAs reduce potential gains from the market? (CATA: None of the above)
What features in RILAs reduce potential losses from market declines? (CATA: None of the
above)
What features in RILAs reduce potential losses from market declines? (CATA: Ceiling)
Yes
83.7%
Yes
85.2%
Yes
87.2%
Question Text
3 years ago, Sally invested $10,000 in a RILA investment product. The investment she
selected is linked to the S&P 500 and has a 3-year investment term. There is a cap rate of
10%, and a buffer of 10%. If the S&P 500 decreased by an average annual rate of 4% per
year, for a total loss of 12.5% over 3 years, how much would Sally’s contract be worth at the
end of the term? (MC)
What features in RILAs reduce potential gains from the market? (CATA: Participation Rate)
If you own a RILA with a 10% floor, you can lose a maximum of 10% over the time that
you are invested in the annuity. (T/F)
What features in RILAs reduce potential losses from market declines? (CATA: Floor)
Note. Question text includes the type of question: True–false questions are labelled (T/F) and choose-all-that-apply questions are
labeled (CATA), with the individual option listed after the colon (e.g., CATA: Cap Rate). Jargon indicates whether the question
used jargon specific to RILAs. The average score is the percentage of all participants who got the question correct.
Investor Testing Report on Registered Index-Linked Annuities | 48
Table 10. Liquidity Sub-score.
Question Text
Jargon
At the end of the investment term, you may withdraw your money without penalty. (T/F)
If she had to pay the highest possible fees, about how much should she expect to receive if
she were to pull out all of her money today? (MC)
If your RILA investment has reached the end of its term, which of the following penalties
could apply if you withdraw your money (CATA: Surrender charges)
To avoid paying penalties on withdrawals, you may need to reinvest your money for several
investment terms. (T/F)
Tori is 51 years old and has had a RILA for 15 years. She most recently contributed money
three years ago. Recently, there was a family emergency, and she needs access to the money
now. (CATA: IVA)
Tori is 51 years old and has had a RILA for 15 years. She most recently contributed money
three years ago. Recently, there was a family emergency, and she needs access to the money
now. (CATA: Tax Penalties)
Surrender charges, interim value adjustments and tax penalties are all waived if the investor
is over age 59.5. (T/F)
An investor who is 85 years old faces no penalties for withdrawing funds. (T/F)
If your RILA investment has reached the end of its term, which of the following penalties
could apply if you withdraw your money (CATA: None of the above)
If the surrender period is 9 years, the surrender charge applies only if you withdraw money
within 9 years of your last payment. (T/F)
Tori is 51 years old and has had a RILA for 15 years. She most recently contributed money
three years ago. Recently, there was a family emergency, and she needs access to the money
now. (CATA: Surrender)
An investor who is 58 years old faces no tax penalty for withdrawing funds. (T/F)
If your RILA investment has reached the end of its term, which of the following penalties
could apply if you withdraw your money (CATA: Transfer fees)
If your RILA investment has reached the end of its term, which of the following penalties
could apply if you withdraw your money (CATA: IVA)
If you invest in a RILA, you can access your money whenever you need it at no cost. (T/F)
Tori is 51 years old and has had a RILA for 15 years. She most recently contributed money
three years ago. Recently, there was a family emergency, and she needs access to the money
now. (CATA: Transfer fees)
If your RILA investment has reached the end of its term, which of the following penalties
could apply if you withdraw your money (CATA: Rollover fees)
Tori is 51 years old and has had a RILA for 15 years. She most recently contributed money
three years ago. Recently, there was a family emergency, and she needs access to the money
now. (CATA: None of the above)
Tori is 51 years old and has had a RILA for 15 years. She most recently contributed money
three years ago. Recently, there was a family emergency, and she needs access to the money
now. (CATA: Rollover fees)
Yes
Average
Score
17.5%
No
20.2%
Yes
31.5%
Yes
36.3%
Yes
45.4%
Yes
51.3%
Yes
55.9%
No
57.1%
Yes
58.4%
Yes
67.8%
Yes
68.4%
No
73.7%
Yes
79.1%
Yes
79.4%
No
80.7%
Yes
85.1%
Yes
87.5%
Yes
93%
Yes
93.5%
Note. Question text includes the type of question: True–false questions are labelled (T/F), multiple choice questions are labelled
(M/C), and choose-all-that-apply questions are labeled (CATA), with the individual option listed after the colon (e.g., CATA: Cap
Rate). Jargon indicates whether the question used jargon specific to RILAs. The average score is the percentage of all
participants who got the question correct.
Investor Testing Report on Registered Index-Linked Annuities | 49
In Table 11, we compare results across treatment arms for overall comprehension scores.
The different introductions to the KIT rows did not have a significant impact on this measure (all
p values > 0.21). Neither the Q&A format nor the bonus payments resulted in significantly
different average comprehension (KIT structure: p = 0.37; bonus payments: p = .24).
Table 11. Summary of Results Across Treatment Arms.
Testing Condition
Introduction
Decision focus
Key terms
Benefits only
Control
KIT Structure
Non-Q&A
Q&A
Bonus Payments
No
Yes
Overall Comprehension
Mean % Correct (SD)
60.2 (13.7)
58.5 (14.3)
58.7 (12.9)
59.1 (12)
58.8 (13.4)
59.4 (13.1)
59.5 (13.4)
58.7 (13)
When splitting out comprehension by the four conceptual sub-scores (see Tables 7–10 for
question classification), we found a few significant differences between the treatment conditions,
when controlling for the other testing factors, but these were relatively small in magnitude (see
Table 12 for condition means). For the Appropriateness sub-score analyses, we changed the
reference category to be the decision focus introduction without the Q&A structure and no
variable incentives. The decision focus introduction was intended to explain the conditions that
would likely make a RILA an appropriate choice and to explain the basic structural features and
key aspects of the contract without using jargon. As such, the other introductions were expected
to lead to worse comprehension of this specific conceptual area. Although participants in the key
terms introduction (4.5 percentage-point decrease; p = 0.005) and the control condition (3.3
percentage-point decrease; p = 0.04) fared significantly worse than those in the decision focus
introduction condition, the difference between the decision focus introduction and the benefitsonly introduction conditions demonstrated only marginally statistically significant differences
(3.1 percentage-point decrease; p = 0.053). These results suggest the decision focus
introduction shows some promise, although the overall levels of comprehension leave room for
improvement. The Q&A versus Non-Q&A variations that were tested on the KIT did not
significantly affect comprehension on the Appropriateness sub-score (0.2 percentage points;
p = 0.88), nor did the variable incentives (p = 0.22).
Investor Testing Report on Registered Index-Linked Annuities | 50
Table 12. Summary of Results by Sub-Score and Treatment Arms.
Testing Condition
Introduction
Decision focus
Key terms
Benefits only
None
KIT Structure
Non-Q&A
Q&A
Bonus Payments
No
Yes
Sub-Score Comprehension (Mean % Correct [SD])
Appropriateness
How a RILA Works
Liquidity
Insurance
60.1 (27.7)
55.5 (28.1)
57.0 (28.1)
56.8 (28.4)
52.9 (20.7)
48.0 (20.5)
48.5 (19.3)
48.3 (18)
62.7 (13.9)
61.1 (15.1)
62.2 (13.9)
62.9 (13.2)
56.6 (15.7)
58.4 (16.3)
56.6 (15.1)
56.9 (15.5)
57.1 (27.9)
57.4 (28.4)
48.4 (19.4)
50.2 (20)
62.1 (14.4)
62.3 (13.7)
56.9 (15.8)
57.3 (15.6)
58.0 (28.2)
56.5 (28.1)
49.9 (19.8)
48.7 (19.6)
62.2 (14.1)
62.2 (14)
57.8 (15.8)
56.4 (15.5)
For the How a RILA Works sub-score, overall comprehension was about 49%, which is
not different from what we would expect if everyone randomly guessed on these questions (one
sample proportion test; p = 0.08). We analyzed comprehension on this sub-score using the
typical analysis structure, where the control introduction was the reference category in the
regression. Using that structure, we again observed higher scores for those who received the
decision focus introduction compared to the control introduction (by 4.4 percentage points;
p < 0.001). There were no differences between the control introduction and the other two
introductions (p values > .72). In addition, participants who received the Q&A format
demonstrated higher scores (by 1.6 percentage points [p = 0.04]) than did participants who
received the Non-Q&A KIT format. Again, the results only represent a slight increase in
comprehension for participants in the decision focus introduction, but the direction of change
demonstrates some promise.
For both the Liquidity and Insurance sub-score, we only saw one significant difference
for each sub-score. The key terms introduction had significantly lower average comprehension
on the Liquidity sub-score (1.8 percentage-point decrease, p = 0.025) compared to the control
introduction. As for the Insurance sub-score, the variable incentive condition had statistically
significantly higher comprehension than the condition without an incentive (1.5 percentage-point
increase, p = 0.02). These small differences were economically very modest improvements
given the absolute comprehension levels on these sub-scales.
One potential concern is that the use of jargon inhibits investors’ ability to understand
RILA investment options and features. To test this hypothesis, we compared how many
questions participants answered correctly when the questions did or did not include jargon. Our
experiment was not designed to test this hypothesis; thus, the use of jargon was not randomized
or balanced within sub-scores, so our results should be considered exploratory analysis. When
pooling the How a RILA Works questions into groups that did or did not use jargon in the
question text, we found that the average percentage correct for the questions without jargon was
Investor Testing Report on Registered Index-Linked Annuities | 51
substantially higher (non-jargon average percentage correct = 57.2%; jargon average percentage
correct = 33.5%; paired one-sample t-test p < 0.001). This finding is suggestive evidence that
jargon may create substantial barriers to comprehension. Additional research on the use of
jargon in disclosures, particularly for such a technical product, is necessary to understand
potential negative effects on comprehension and associated remedies (see Chin et al., 2023 for
further discussion).
Subgroup Analysis
Given the complexity of the RILA products, it is reasonable to believe that more
sophisticated investors might fare better in terms of comprehension, since they may have more
background knowledge that could help contextualize the disclosed information. As such, we
conducted subgroup analysis for four subgroups that were defined according to subjective
financial literacy levels, investor status, annuity ownership, and our initial recruitment groups.
These analyses were performed by interacting indicator variables for subgroups with the testing
factors.
Subjective Financial Literacy
Three questions on a 5-point scale that captured subjective financial literacy were
summed together (self-assessed ability to deal with day-to-day financial matters, self-assessed
math ability, and self-assessed proficiency in dealing with saving and investing financial
matters), creating a score that ranged from 3 to 15. 21 There was a positive correlation between
subjective financial literacy and overall comprehension. On average, a 1-point increase in
subjective financial literacy was associated with a 1.3 percentage-point increase in overall
comprehension (p < 0.001). In other words, for the lowest score, we would predict the average
comprehension to be 47%, whereas at the median of the scale (12), we would expect average
comprehension to be 59%. We did not find that treatment effects varied across conditions based
on subjective financial literacy. 22
Investor Status
Participants were classified into three groups: those with no investments in stocks, bonds,
mutual funds, or other securities (“Non-Investors”); those with investments exclusively in
retirement savings accounts (“Retirement Only”); and those with investments outside of
retirement accounts (“Independent Investors”). 23 In prior research, these categories have been
associated with increasing financial sophistication, including investment knowledge and trading
behavior (Chin et al., 2022a; Chin et al., 2023).
21
There were five people in the sample who did not respond to at least one of the three subjective financial literacy questions, so we could not
calculate a sum score for them; thus, they were excluded from the moderation analysis.
22
Subjective financial literacy was mean centered and interacted with indicator variables for each of the treatment conditions. For example, there
was an estimated coefficient for the key terms introduction, subjective financial literacy, and the interaction between the key terms introduction
and subjective financial literacy. Interactions between the different testing factors were not included.
23
Note that by definition, the Independent Investors group consisted of individuals who may or may not have retirement savings accounts. This
is consistent with the definition used in Chin et al., (2021) and Chin et al., (2022).
Investor Testing Report on Registered Index-Linked Annuities | 52
In the current context, investor types are relevant for two reasons: First, Independent
Investors likely have more knowledge about investing compared to members of the other groups,
which may make it easier for them to understand the structure and features of RILAs.
Retirement-Only Investors likely have more knowledge about investing than Non-Investors, but
less knowledge than Independent Investors. The retirement savings design of RILAs is also
likely to be more familiar to those with retirement savings (Retirement-Only and some
Independent Investors). Second, Independent and Retirement-Only Investors are more likely
than Non-Investors to purchase these products given that they have financial assets (by
definition), and there are minimum purchase amounts required for RILAs. From an investor
protection standpoint, it is relevant to understand how comprehension may vary by investor type,
including when viewing different disclosures.
On average, Independent Investors had a higher overall comprehension, with an average
score of 61%, which is 3.4 percentage points higher than the average for the Retirement-Only
Investors (p < 0.001, not controlling for testing factors) and 11.7 percentage points higher than
the average for the Non-Investors (p < 0.001, not controlling for testing factors). As for
moderation of the testing effects by investor status, we only saw one significant interaction
between investor status and the impact of conditions. The effect of the Q&A KIT structure on
overall comprehension (vs. Non-Q&A structure) was larger for certain groups, estimated at a 0.5
percentage-point increase for Independent Investors versus a 5.7 percentage-point increase for
Non-Investors (p = 0.011). None of the other differences by experimental condition were
significant, nor were the effects different between the Independent Investors and the other two
groups (Retirement-Only Investors and Non-Investors). Given the number of statistical tests in
the moderation analyses, this difference would likely not remain significant if we controlled for
the number of statistical tests conducted.
Annuity Ownership
Participants were classified into two groups: those with annuities and those without
annuities. There was a small but statistically significant difference in the average overall
comprehension, with annuity owners scoring 2.6 percentage points higher on overall
comprehension (p < 0.001). We did not see any variation in the effects of the experimental
conditions for participants of different annuity ownership.
Time
As exploratory analysis, total time spent on the survey and time spent on the four KIT
rows were compared across the experimental conditions; 24 this analysis helps to understand if the
introduction and KIT experimental conditions affected efficiency and engagement. The median
time spent on the KIT rows was 87 seconds, or just shy of one and half minutes, whereas the
Given the distribution of time spent on the whole survey and on all KIT rows, we log-transformed the time outcome (using natural log) to
better satisfy the modeling assumptions of the OLS regression.
24
Investor Testing Report on Registered Index-Linked Annuities | 53
median total survey time was 12.4 minutes. Both time spent on the KIT rows and time spent on
the entire survey were positively correlated with overall comprehension (Figure 15).
The amount of time spent on the KIT was highly skewed with a median of 87 seconds,
but a max over 23 minutes. 25 As a result, in our analysis we focus on the median and the log of
KIT time, and in the graph below we use a log scale on the x-axis; we used a simple least squares
regression to describe the relationship between time spent and the log of the percent of questions
correct (the log of KIT time better accounts for the dispersion in values). An individual who
spent the median amount of time on the KIT rows (87 seconds) is associated with an average
score of 59.7% of the comprehension questions correct. Someone spending an additional 10
seconds on the KIT rows (97 seconds), is associated with an average score of 60.4% of the
questions correct. As for total survey time, having spent the median amount of time (12.4
minutes), is associated with an average score of 59.1% of the overall comprehension questions
correct. Spending 1 minute longer on the entire survey is associated with an average
comprehension score of 59.8% of the questions correct.
Figure 15. Relationship Between Transformed KIT Time and Comprehension.
Note. Figure 15 shows the relationship between time spent on all KIT rows and overall comprehension. The time spent on all
KIT rows was log-transformed and then relabeled with the actual number of seconds someone spent on the KIT row pages. The
blue line is the ordinary least squares regression of the variables plotted.
There was no statistically significant difference across the experimental conditions
(varying the introduction or KIT format) in the total amount of time spent completing the survey
Given the online format of survey administration, it is possible that the longer survey response times do not reflect active engagement. For
example, a participant could start a survey and pause to run a household errand. Trimming of observations was considered, but not pursued here
given that our evidence on timing is purely exploratory. The log transformation helps to deal with some of the resulting time dispersion.
25
Investor Testing Report on Registered Index-Linked Annuities | 54
or the total combined time spent in the survey reviewing the respective introduction and the KIT
rows together (all p values > 0.1). However, time spent on the KIT rows alone (omitting time
spent on introductions) varied somewhat. Compared to those in the control introduction
condition, participants who saw the decision focus introduction spent less time on the four KIT
rows (b = -.32, p < 0.001). Similarly, the key terms and benefits-only introductions led to
significantly less average time spent on the four KIT rows (key terms: b = -.24, p < 0.001,
benefits only: b = -.16, p = 0.002). Translating these values into seconds, the average time spent
on the KIT rows in the control introduction condition was about 134 seconds. For the decision
focus introduction, the average time spent on all KIT rows was approximately 102 seconds. The
average time spent on all KIT rows was 112 seconds for the key terms introduction and 121
seconds the benefits-only introduction. Concretely, these results overall mean that on average,
the reduction in time spent on the KIT rows for the non-control introductions was reallocated to
time spent on the introductory text.
These results linking time spent and comprehension are descriptive in nature and not
intended for causal interpretation. Nevertheless, they suggest an interest in this relationship for
further study as there are at least two interpretations that are consistent with our observations.
First, it is possible that the introductory materials allowed participants to more efficiently process
subsequent information about the RILAs, accelerating reading of the information presented in
the KIT rows with limited detrimental effects. On the contrary, the introductions, which provide
an alternative source of some of the KIT information, could have simply increased participants’
subjective sense of familiarity, leading them to speed through the KIT while retaining less
information presented in the KIT. Additional research is needed to identify how common either
of these effects may be and what kinds of introductions are more likely to result in better
comprehension.
Summary of Quantitative Testing
The quantitative testing suggests that understanding RILAs is challenging for many
consumers. Overall comprehension was low, and there were no statistically significant
differences in the percentage of correct responses for overall comprehension across treatment
arms. In terms of our sub-score concepts, the decision focus introduction variant demonstrated
some promise in terms of basic RILA concepts and appropriateness, but it did not affect overall
scores or scores on the Liquidity and Insurance sub-score. The Q&A KIT format demonstrated a
statistically significant, albeit quantitatively small, improvement over the Non-Q&A KIT format.
Finally, more sophisticated respondents demonstrated higher (though still arguably low) levels of
comprehension. Even with the most promising variants of the introduction and the KIT (the
decision focus introduction and the Q&A KIT), there were only modest improvements, and no
variation in treatment effects across subgroups. We are unaware of any disclosure that has
achieved perfect levels of comprehension in quantitative testing. However, it is possible that
additional variants could further increase investors’ comprehension and decision-making quality.
Investor Testing Report on Registered Index-Linked Annuities | 55
Jargon demonstrated a clear barrier in our quantitative testing. We found that
comprehension was substantially lower for our jargon-laden questions relative to the non-jargon
questions. Yet, even the questions without jargon yielded average comprehension levels of at
most 65%. This suggests that substantial conceptual barriers surround RILAs. Some
comprehension questions highlight how hard these products were to understand. For example,
only 12.6% of participants correctly identified that RILAs are investment vehicles that are
intended purely as retirement savings vehicles rather than intended for achieving other goals
(e.g., education, home purchasing). This question was intended to be among our most basic
questions about the products. 26 From the qualitative results, we found that people had trouble
understanding what features apply to the overall contract versus the individual RILA investment.
This pattern was confirmed, as only 21% of participants correctly identified that a floor applies
to a RILA investment option rather than to the insurance contract. 27
To the extent that disclosure is the policy instrument of choice for addressing investor
protection concerns, future research may choose to build on the potential merits of the decision
focus introduction condition (and, to a limited extent, the Q&A KIT format). In particular, as
the decision focus introduction was designed to provide decision-relevant information up front,
further research could seek to understand what factors are most important to the actual decisions
the investor needs to make (e.g., Should I invest in a RILA?), followed by designing a disclosure
that is organized with that structure in mind. If pursued, these materials should also follow other
principles of good communications (e.g., Bruine de Bruin & Bostrom, 2013; Sunstein, 2010) to
avoid information overload (Scheibehenne, 2010). At the same time, the fact that these products
are typically professionally sold and highly complex suggests that policy attention to investment
professionals’ conduct and marketing standards may merit consideration, at least as a
complement to disclosure.
On this item, an OLS regression model showed that comprehension was significantly higher for the decision focus introduction condition
relative to the control introduction condition (a difference of 5.9 percentage points, p = 0.002)
27
There were no differences between the conditions on this item.
26
Investor Testing Report on Registered Index-Linked Annuities | 56
7. Conclusions
Summary of Methods
This report presents the results of research and investor testing that fulfills the 2022
Congressional directive to “engage in investor testing” regarding RILAs. 28 In particular, we
conducted (1) a simulation analysis to understand how RILA products and their insurance
features affect investment results and potential investor outcomes when no early withdrawals are
made, (2) a review of common marketing practices, (3) qualitative testing, and (4) quantitative
testing. By examining RILAs from multiple angles, we provide more holistic, multidisciplinary
insight regarding these products.
In our simulation analysis, we measured the benefits and risks of hypothetical RILA
products over different time periods, and we measured how highly complex sets of insurance
features would have operated under various market environments. These simulations enabled us
to estimate purchaser outcomes for up to 3 decades of market history, using current product
offerings, even though we were unable to access direct information on actual investor outcomes
and true historical RILA insurance features. Part of the simulation work concentrated on
commonly offered insurance features and on how variants of those features could affect
outcomes. Overall, the simulation results are crucial to understanding the economic features and
economic value of RILA products—particularly performance and risk—that might be worth
consideration in product disclosure. We identified investment options that increase certain risk
measures in historical simulations despite the products largely being sold as risk mitigation
instruments, raising concerns that some investment options may not operate as consumers
expect. More investigation along these lines and additional transparency is needed to fully
understand RILAs and how the market evolves over time.
In our review of product providers’ common marketing practices, we explored the
advertised benefits of RILAs for potential investors. This component of our research set the
stage for our qualitative and quantitative testing and helped provide basic information about how
the products are presented to potential purchasers.
In our qualitative testing, using an in-depth interview approach with 20 annuity owners
and potential owners, we explored familiarity with RILAs and how consumers approach the
terminology and features of RILA products. These interviews provided initial information on
how well consumers understood the RILA disclosures and applied the information they received
to the scenarios. The open-ended approach also allowed participants to raise questions about the
materials, bringing up issues that were relevant for them. Ultimately, the qualitative study was
crucial to our ability to design our quantitative testing.
28
See Section 1 of this report for more information.
Investor Testing Report on Registered Index-Linked Annuities | 57
Finally, the quantitative testing allowed us to formally test certain variants of the
disclosure formats and introductory texts. The testing, which employed a nonprobability sample,
provided additional information on the ability of these variants to affect investors’ understanding
of the products and their ability to apply knowledge of the products.
Summary of Findings and Discussion
Each of our research efforts yielded insight into RILA products or disclosure.
Simulations. In our exploration of the insurance features, we identified that, as expected,
certain insurance features (such as caps) can, in some cases, limit returns relative to the
underlying index and that downside protections may avert relative losses. However, the benefits
of such risk reduction should not be overstated: An identical set of insurance features was
associated with a wide variety of investment outcomes (and relative outcomes), depending on
when the investment occurred, due to the timing risk associated with RILA investing. We also
identified some instances in which RILA options would have increased both the probability and
magnitude of loss, while substantially reducing the upside gain.
The simulations helped highlight the conditions under which relative gains and losses
occur. Despite the fact that many of the insurance features appear to operate mechanically (e.g.,
caps limit returns), it is difficult to fully understand the implications of the insurance features.
The insurance features only activate under specific index movements, and the time-limited
investment terms mean that gains or losses over a prolonged period of time may not be the same
as a series of repeated investments that span the same period. For instance, a RILA contract
consisting of two 3-year investment options may differ from a single 6-year investment option
over the same period. In the first case (with two 3-year terms), the relative value is determined
by the index movements halfway through the 6-year period, since a RILA with a 3-year term
would have returns credited or removed after each investment term. This midstream evaluation
of returns means that the ultimate value may be higher, lower, or the same.
Marketing Practices. Overall, our review of RILA marketing materials suggests that the
number of choices an investor needs to make when investing in a RILA and when selecting
among available investment options (e.g., choices of index, buffer, floor, cap rate, participation
rate, term length) could easily be overwhelming. There are many individual decisions to make
with RILA products and while the promotional materials focus on the highly important decision
of whether an individual should consider a RILA, they often do not include information that
could aid decisions about which specific options to choose within that RILA. This complexity
might increase investors’ dependence on other sources of information, including professionals to
identify whether a RILA is appropriate for them, to explain the pros and cons of the different
menu of investment options, and to aid the investor in determining which combination of
features is likely to be in their best interests. However, Scholl et al. (2018) report that fewer than
half of investors report using professional financial advice.
Investor Testing Report on Registered Index-Linked Annuities | 58
Qualitative Testing. We identified a range of barriers to investors’ understanding of
RILAs and associated disclosures. In our study, one recurring issue was the terminology used to
describe RILA contracts and the investment options. The jargon was confusing and led to
incorrect beliefs for some study participants. For example, many participants interpreted the
phrase “term” or “investment term” largely to mean the length of time in which they are locked
into the RILA contract, whereas it actually refers to the length of time they are locked into a
particular investment option. Use of this terminology led many participants to the wrong
impression about how other features of the product work and misdirected interpretations of the
disclosures about those features; for example, the structural distinction between the RILA
contract itself (for which surrender charges and tax penalties may apply for withdrawing money
even after a term has completed) and the investment options (for which an early withdrawal
would additionally require payment of an Interim Value Adjustment) remained unclear.
There was significant confusion about the features and expenses associated with RILAs,
including concepts like the buffer and the Interim Value Adjustment. Our participant pool
targeted annuity owners and shoppers and had several highly sophisticated investors. Even these
individuals demonstrated confusion or lost interest in RILAs once that confusion was abated.
Several variants of the disclosures were examined, including variations on the KIT as
well as inclusion of graphics or tables to illustrate several key points, but each of these variants
pointed to the difficulty in communicating key features of RILAs to the investing public,
including the communication of basic aspects such as how the products work and what purpose
they are intended to serve for the investor. The ineffectiveness of the graphical and tabular
information we presented may be considered specific to the graphical and tabular designs we
used: we did not study whether the graphics or tabular presentations in general could be
effective, which was not possible given the limited time available for this research; such testing
would require examining many different types of each display.
Quantitative Testing. Our testing examined a set of KIT variations, different
introductions to the RILA products, and the role of variable incentives in affecting
comprehension. We largely observed that there were no significant differences, on average,
among any of the treatment conditions. Even when we found significant differences, their
magnitudes were modest (e.g., 1 to 5 percentage points), particularly considering the average
comprehension levels observed, with the decision focus introduction showing promise and the
Q&A KIT format having a slight improvement in certain sub-scores. Certain subgroups of
participants varied in their comprehension but not differentially across treatment conditions. Our
results do not indicate that disclosures writ large are ineffective at communicating relevant
information to investors. Rather, they suggest that the range of options we tested did not
differentially impact comprehension. Moreover, our results suggest that misunderstanding
permeated a variety of conceptual issues regarding RILAs. There were relatively low levels of
comprehension regarding both technically difficult questions about the insurance features and
basic questions regarding the purpose of a RILA.
Investor Testing Report on Registered Index-Linked Annuities | 59
Although our variations on the KIT were somewhat limited and a broader range of
options could be explored, the limited effects of the alternative introductory texts suggest that
disclosure variants may not be a complete solution for investors who are interested in RILAs.
We provided participants with one of four introductory texts (no introduction, a benefits-only
introduction, and two substantive introductions). These different introductions did not result in
improved participant outcomes on overall comprehension despite a wide variation in not only
clarity, but also in the information provided. Along certain sub-score dimensions, the decision
focus introduction demonstrated the most promise. Nevertheless, further pursuit of alternatives
may be worth consideration.
Implications of the Research: The Economics of RILAs
To make an informed decision about purchasing a RILA, investors need to have a full
understanding of the economics of RILAs—the costs, benefits, and how future behavior may
affect subsequent outcomes. By understanding the aspects of RILAs that most affect investor
outcomes, policy activities can be more targeted. Our research suggests that RILAs are highly
complex financial instruments with features that are difficult to understand. Investors may need
additional information and assistance to understand these products’ likely financial outcomes. In
some cases, the relevant information is lacking in extant documents. Further testing may be
warranted to determine the best ways to communicate with investors:
•
•
A clearer distinction between the contract and the investment term: Many of our
study participants reported that it is difficult to reconcile the 1-, 3-, and 6-year terms with
the statement that a RILA is “a long-term investment product.” Many assumed that their
funds could be liquidated without penalty at the end of a 1-year term, and many had
different subjective interpretations of the period implied by “long term.” Information
about the typical or anticipated contract length and the minimum contract duration to
withdraw money without penalty could be helpful for investors to understand the
distinction between the contract and the investment option term.
The financial implications of caps, floors, buffers, and other insurance features:
Insurance features are particularly confusing when a RILA contract contains multiple
investment options that are renewed at the end of the investment term. To better illustrate
those features, investors may benefit from information about the relationship between
premium payments (invested) and money out (e.g., the investment balance or funds
distributed) and about how those relationships vary in different market conditions.
Relatedly, an understanding of how a given insurance feature works in a single point-topoint investment term does not imply an understanding over the life of an investor’s
RILA contract (due to value recalculation at the end of an investment term, the potential
reinvestment at the end of the investment term, and the repeated timing risk from these
adjustments). Better illustration of these relationships may be necessary.
Investor Testing Report on Registered Index-Linked Annuities | 60
•
•
•
•
Information on investment alternatives: Investors may benefit from seeing comparative
information about hypothetical investments in the underlying index, a plausible
alternative to investing in a RILA. In previous research, investment-related attributes
have been compared to performance benchmarks (Chin et al., 2022a) or the market
overall (Scholl, Craig, & Chin, in press) to provide context. However, without future
research, we cannot determine whether benefits from comparisons would outweigh other
potential costs, such as those stemming from information overload.
The financial consequences of early withdrawals: Given the substantial penalties
associated with early withdrawals (i.e., Interim Value Adjustments, surrender charges,
and penalties), additional consideration should be given to the presentation of quantifying
these costs and how they affect investment value. Investors may be interested in the
extent to which others have made early withdrawals and how a given RILA balance could
change when the investment is withdrawn mid-term.
Information on risks: Although the hypothetical KIT discussed risks, the descriptions
did not always provide information on the likelihood and magnitude of the risks. We
believe a more direct and decision-relevant discussion of risk merits consideration,
especially because RILA advertisements prominently mention risk mitigation. Financial
models may concentrate on the standard deviation of returns or some function thereof
(e.g., the Sharpe ratio), but risk in the common vernacular may be quite different
(Holzmeister et al., 2020) and may include the chances of any loss, the chances of a loss
greater than 25%, or some other concept. In our simulations, we found that over 6- and
12-year investment durations, some of the indexes underlying the RILAs rarely lost
money, and when they did, the losses tended to be small in magnitude. Some of the
RILA insurance options we explored mitigated these losses at the cost of considerable
upside, raising questions about whether investors are fully informed about returns. Some
examples exacerbated risk in the sense of certain risk definitions. Bearing this in mind,
the historical probability of a negative return on a RILA product versus the historical
probability of a negative return in the index over the lifetime of a RILA contract might be
informative. Further research could continue to explore how investors conceive of risks
and how such risks can be effectively communicated (for existing research, see e.g.,
Holzmeister et al., 2020; Morgan et al., 2002; Okan, Stone, & Bruine de Bruin, 2018).
Informa
This text is long and has been trimmed here. Open the source document for the complete record.
This is a copy of a public record, reproduced as it was published. It is not legal advice, and it may not be the version a court would rely on. Check the official source before you cite it.