(AGENPARL) – mar 04 febbraio 2025 Questioni di Economia e Finanza
(Occasional Papers)
The cost differential between unit-linked policies
and mutual funds
Number
February 2025
by Angelo Nunnari and Agostino Tripodi
Questioni di Economia e Finanza
(Occasional Papers)
The cost differential between unit-linked policies
and mutual funds
by Angelo Nunnari and Agostino Tripodi
Number 907 – February 2025
The series Occasional Papers presents studies and documents on issues pertaining to
the institutional tasks of the Bank of Italy and the Eurosystem. The Occasional Papers appear
alongside the Working Papers series which are specifically aimed at providing original contributions
to economic research.
The Occasional Papers include studies conducted within the Bank of Italy, sometimes
in cooperation with the Eurosystem or other institutions. The views expressed in the studies are those of
the authors and do not involve the responsibility of the institutions to which they belong.
The series is available online at http://www.bancaditalia.it .
ISSN 1972-6643 (online)
Designed by the Printing and Publishing Division of the Bank of Italy
THE COST DIFFERENTIAL BETWEEN UNIT-LINKED POLICIES
AND MUTUAL FUNDS
by Angelo Nunnari* and Agostino Tripodi**
Abstract
This study estimates the cost differential between investing in unit-linked insurance policies
and investing directly in the same underlying mutual funds, using information from product
prospectuses. The cost of investment is measured at the level of the single underlying
investment fund by the Reduction in Yield (RIY). Using Monte Carlo simulations, the analysis
controls for factors such as fund type, investment amount, holding period and investor age. The
main findings indicate that the simulated cost of unit-linked policies is generally significantly
higher than that of mutual funds purchased on the retail market, mainly due to higher ongoing
costs. The cost differential ranges from 0.5 to 2.5 per cent and is higher when investing in
money market funds, ETFs, and policies distributed by financial advisors. Our findings hold
when controlling for the actuarial component of insurance policies, possible policy rebates, the
costs of portfolio reallocation and the actual portfolio allocation of insurance companies. The
extra costs of unit-linked policies may be justified by advisory services and other additional
features that insurance companies may offer. It could also reflect information and cost
transparency issues or limited competition in the distribution of savings products.
JEL Classification: D14, G22, L11.
Keywords: unit-linked insurance policies, investment funds, cost comparison, households
investments, Monte Carlo simulations.
DOI: 10.32057/0.QEF.2025.907
________________________________________
* Bank of Italy, Directorate General for Economics, Statistics and Research – Financial Stability Directorate.
** IVASS, Research and Data Management Directorate, Research and Statistics Division.
1. Introduction and main results1
In the last 15 years, Italian households increased significantly their exposure to financial markets mainly
by investing in asset management products. In particular, household investments in insurance life policies
(with-profit or unit-linked) and investment funds as a percentage of their financial wealth rose
respectively from 10.8 and 8.6 per cent in 2007 to 14.4 and 12.8 per cent in December 2023). Premiums
yearly underwritten of unit-linked insurance policies (life products whose investment risk is borne directly
by the underwriter) reached a peak of 44 billion euros in 2021 after a prolonged period of very low
interest rates; since the second half 2022, following the start of the tightening cycle of monetary policy,
their popularity partially decreased, as returns from treasury bonds and traditional insurance products
became more attractive.
Despite the widespread diffusion of managed savings products, retail investors may not always be able
to assess the impact of their costs on investment returns. Specifically, the cost borne by a household for
holding an insurance product or an investment fund varies greatly across contracts2 as it depends on
several types of charges and the specific characteristics of the investment. This raises a number of
regulatory issues regarding product transparency, consumer protection and the degree of competition in
the insurance industry3. Moreover, given the close similarity of unit-linked products and mutual funds,
the relative cost competitiveness of the two instruments is an important factor for the development of
their markets.
In this study we compare the costs of a sample of unit linked insurance policies with those of equivalent
investment funds. In order to make the exercise as clean as possible we focus only on unit linked policies
that invest in external funds (i.e., funds also available for investment on the retail market). In this case, in
fact, a retail investor faces two financially equivalent options: underwriting a unit-linked policy whose
capital is invested in a specific mutual fund or, alternatively, directly investing in the retail shares of that
same fund. This allow us to compare their costs by controlling for the financial characteristics of each
product.
For each instrument (insurance policy or investment find) we simulate the impact of all the main fees
charged by both asset managers and distributors (based on information extracted from products’
prospectuses) on the investment return (“reduction in yield”, RIY). We then compute the differential
between the “reduction in yield” of an investment in an insurance policy in our sample and that of the
corresponding retail fund. We use a Monte Carlo simulation to properly account for the stochastic nature
of returns, controlling for type of fund, amount invested, holding period and age of the hypothetical
investor. We then analyze the distribution of the simulated cost differential across funds, policies and
companies. Finally, we investigate which product characteristics are associated with an increase in the
cost differential and which types of fees contribute the most to it.
We find that on average the simulated cost differential between a unit linked policy and the underlying
investment funds is positive, with the first and the third quartile equal to 1.0 and 1.7 per cent respectively.
Most of the differential is explained by the fact that insurance products have higher ongoing costs than
mutual funds. Initial costs (paid upfront by investors) are on average more expensive for mutual funds
views expressed in this paper are those of the authors and do not reflect those of the Bank of Italy and IVASS. We thank
for the helpful comments and suggestions Riccardo Cesari, Marco Cosconati, Antonio Rosario De Pascalis, Stefano De Polis,
Alessio De Vincenzo, Giovanni Guazzarotti, Sabrina Pastorelli and all the participants at the Comitato IVASS-Banca d’Italia
meetings. We also thank Lanfranco Lunghi who participated in the early stages of this work.
2 According to ESMA (Annual report, 2023) in EU the charges of UCITS ranges from an average of 0.9 for bond funds to an
average of 2.6 for Real Estate funds. According to EIOPA (Costs and Past Performance Report, EIOPA, December 2023), the
cost of Italian unit-linked products ranges from about 1 per cent to above 5 per cent. ESMA and EIOPA estimates cannot
be directly compared as the underlying hypothesis are not homogeneous.
3 Cfr. Luigi Federico Signorini, speech at ANIA annual assembly, 4 July 2023.
than for unit linked policies, but their incidence over total costs is contained. The simulated cost
differential varies greatly depending on the characteristics of the contract (investment amount, holding
period and age of the policy underwriter) and the type of funds the policy premium is invested in. For
example, the cost differential is higher for policies invested in money market funds (because lower-risk
funds have generally lower fees), for exchange traded funds (since ETFs do not apply initial costs on
retail investors), and for insurance policies distributed by financial advisors (whose distribution fees are
on average higher than those distributed by banks). The cost differential is instead lower when policies
are invested funds’ classes reserved to institutional investors. Our results are confirmed when controlling
for the actuarial component of insurance policies, possible policy rebates and costs of portfolio
reallocation. We also compute a volume-weighted average cost differential based on company unit linked
portfolios and show that the main results are robust when accounting for actual investment choices.
Our simulation of the cost differential is based on the information included in the products’ prospectuses
and on specific hypotheses about the characteristics of the individual contracts actually signed (that we
do not observe). We also assume that the investor can replicate the investment portfolio of the unit linked
policy (buying the same funds) without having to bear any extra costs of advisory or security custody
account services. Our analysis shows how the distribution of the differential across funds varies under
different hypothesis and possible investment choices and it is not meant to provide a point estimate of
the actual cost differential borne by retail investors.
Our results show that under specific circumstances the extra costs of unit-linked policies can be quite
significant; however, we are not able to assess the implication of this evidence on the welfare of retail
investors (since we do not dispose of individual, contract-level data). On the one side, the extra cost
borne by the retail investor can be seen as a compensation for the implicit advisory services offered by
the insurance company or for the additional features eventually offered by the insurance contract (e.g.
death coverage, non-seizability of the capital invested, exemption from the inheritance tax, etc.). On the
other side, it can reflect a certain degree of stickiness in investors’ behavior (possibly linked to demand
complementarities or the fact that retail investors are not able to properly assess and compare the costs
different saving products) and its impact in terms of reduced industry competition. In the latter case,
policies fostering information and cost transparency or incentivizing competition in the distribution of
saving products might contribute to a reduction of costs for the retail investor.
The paper develops as follows. In section 2 we illustrate the cost structure of unit linked policies and
mutual funds, discussing their differences and their similarities. In sections 3 and 4, we describe the
dataset and the methodology used in the analysis. In sections 5 and 6 we discuss our main results and
some robustness checks.
2. Unit-linked policies and mutual funds: markets and products
Our analyses focus on the segment of the unit linked market, where underwriters invest in external
investment funds that are also available for sale in the retail market. Since unit linked policies and funds
are similar retail investment products, in principle we should not find any significant difference in costs
once we control for other institutional or contractual characteristics.
In the last eight years written premium of unit-linked policies in Italy cumulated at 258 billion euros (31%
of global written life premium), while retail net investments in funds were 193 billion. The share of
policies with external funds out of total premium of unit-linked products increased in the last years to
approximately 50 per cent. In terms of costs, in 2023 the average RIY of unit-linked products with
external funds was 3.4 percent as opposed with 2.6 percent for those with internal funds and 1.7 per cent
for traditional policies4.
Billion euros
Billion euros
Figure 1: Premium underwritten of unit-linked products
(billion euros; percentage)
(a) premium underwritten
(b) share of external funds within unit-linked contracts
and share of unit-linked contracts
Global written life premium
Unit-linked premium (1)
share of Unit-linked (1) (2)
share of UL with external funds (2)
Note: (1) It comprehends both unit-linked sold as stand-alone and as hybrid products. (2) Right scale
Source: IVASS
2.1 The cost structure of unit-linked policies and investment funds
The cost structure is quite similar for both products, although the jargon and the components that they
remunerate may be different. We can summarize it in three main steps of value extraction: initial, ongoing
and exit fees. For more details on the single components see the Annex A.1 Costs.
Table 1: Main fees and bonuses by type of product
Initial charges
Ongoing charges
Exit charges
Other charges or
bonuses
Unit-linked policies
Policy fee
Loading cost
Annual management charges1Ongoing
fund’s charges
Surrender charge
Mutual funds
Service fees at entry
Entry charges
Ongoing fund’s charges
Exit charges
Service fees at exit
Coverage fee2 (either within ongoing or
up-front fees)
Initial bonus
Death bonus
Note: (1) Annual management charges (in Italian, costi di gestione e salvaguardia) do not remunerate the activity of asset management but the
maintenance of the list of available funds on behalf of the underwriter and the liquidation of funds which may undergo extraordinary processes, as
for instance asset freezing or block of surrenders. (2) Fees for the death benefit coverage of unit-linked products that may be included either in the
loading cost or in the annual management charges.
2.2 Similarities between unit-linked policies and mutual funds
Unit-linked policies and mutual funds have many similarities. Apart from the death benefits coverage
provided by the insurance policy, both are popular collective retail saving products delegating a
management company and allowing clients to benefit from the economies of specialization and
diversification. In the case of unit-linked policies with external funds, a retail investor could in principle
Data from the Fairmat database
actually achieve the same financial objective by either investing in a unit linked policy or alternatively in
the funds linked to the policy (through a bank, a financial advisor, or an online platform).
The two products are therefore similar under the following three main aspects.
Autonomy of the investors. In both cases, the investor selects autonomously the funds where to invest
(assisted or not by a financial advisor). The list of funds available within an insurance product is similar
to the catalogue of mutual funds supplied by intermediaries.
Freedom in selling or retaining quotes. In both cases, the investment can be redeemed or reallocated at almost
any moment. The only additional limitation for insurance clients is the usual presence of a minimum
holding period of one year.
Similar way of investing. In both cases, the actual asset allocation of the underlying funds is delegated to a
professional management company which must follow a transparent regulation and a pre-defined
management style.
2.3 Advantages of holding a unit-linked policy with respect to an investment fund
Policies may offer some advantages respect to mutual-funds that one should consider when comparing
investments costs. First, a policy may offer the opportunity to invest in funds different from those offered
by a specific bank or personal advisor. Furthermore, under the Italian law, the capital invested in an
insurance product is non-seizable in the event of insolvency, it is not attachable from the authorities and
it is exempt from the succession tax5 upon death of the underwriter. Policies can eventually offer
additional services as active management mechanisms or forms of financial protection6.
2.4 Assumptions and limitations of the analysis
There are several caveats that one needs to be aware of when interpreting the results of our analysis. In
particular, we need to rely on specific assumptions as we do not have real world information at a contract
level.
Specifically:
We assume that the investment in mutual funds is conducted either directly or through an
intermediary (e.g. a bank) but without the assistance of a financial advisory. Financial advice is a
service acquired and paid for independently from other investment services by the retail investor;
it is not widespread in the Italian market. Symmetrically, in the case of a unit linked policy we
assume that the underwriter does not apply for any of the insurance contract additional financial
options comparable to a financial advisory service, e.g. the automatic management of the
investment7.
We do not consider the impact of the inheritance tax exemption of policies upon death of the
insured8. The fiscal treatment of investment income is the same for both unit linked and funds
and it is not considered in the analysis9.
We assume that the client is able to replicate the investment portfolio of the policy (buying the
same funds) and she/he has already opted for a security custody account in the bank which is
needed to buy any financial instrument other than a plain bank deposit. The ongoing costs of a
It is worth noting that inherited wealth, of any kind, under the franchise threshold (1 milion euros) on each heir is always
exempt from the respective tax.
6 These services are offered at the cost of an additional fee that we subtract from the costs of the unit-linked policy.
7 Several policies offer the possibility to opt for premium services as the Stop Loss, Take Profit, automatic rebalancing.
8 However, the benefit starts with a total wealth above 1,000,000 and the maximum level of premium we use in the simulation
is 500,000 euros.
9 In an “administrate savings regime” they both pay 26 percent of the profits and the 0.2 percent of recurring stamp duty.
custody account (e.g. periodical reporting costs, dividend payment fees, etc.) are heterogeneous
across intermediaries (sometimes it is supplied for free).
On a precautionary basis, we set the entry costs for mutual funds at the highest level of the
interval reported in funds’ prospectuses, as we do not have information on the actual costs
applied (as they depend on the specific agreement between the asset management company and
the distributor).
Over-performance fees applied by both insurers and funds are not considered as their net impact
on the differential tends to be zero. In fact, unit-linked policies based on external funds do not
set their specific over-performance fee as the allocation of the underlying funds is done by a third
party, the management company. The latter tends to set them at the same level across all the
classes (either institutional or retail).
In the simulation we exogenously set the holding period of the investment, the age of the
investor/underwriter and the parameters of the returns distribution in order to obtain a ceteris
paribus comparison between the two form of investment.
Apart from the death benefit coverage, we exclude any additional insurance coverage (either
compulsory or discretionary). These insurance coverages carry additional costs that we coherently
do not include in the analysis.
We assume that the investor selects only one investment option at a time, as we do not have data
on the actual investment choices, and we compute a cost differential for each policy-fund
combination. As a robustness test we compute a weighted average cost differential based on the
actual company-specific aggregate portfolios underlying the unit linked products.
We assume that the investor/underwriter never rebalance her/his portfolio, but in a robustness
check we control for the impact on our results of a hypothetical number of portfolio switches.
In our base simulation we do not consider rebates agreed by the insurer and the fund’s
management company on a case by case basis as we do not have this contract level information;
however, we control for a hypothetical level of rebates as a robustness check.
3. Data
3.1 Data sources
The list of all active single premium unit-linked policies with external funds and for each of them the list
of funds available for the underwriters have been extracted from the Fairmat database. The rest of the
characteristics of the insurance products (in particular, the level of each type of fee) have been obtained
by reading contract conditions.
Relevant information on mutual funds (in particular, the level of each type of fee) has been retrieved
through the Morningstar Direct database. Prospectuses issued by the management companies of the
funds were used for identifying for each fund the share classes sold respectively to retail and institutional
investors.
3.2 Descriptive statistics
The sample analyzed includes all the 64 single premium, unit linked policies with external funds
distributed in the Italian market in November 2022 and issued by companies still operating at the end of
2023. We excluded 22 products issued by two foreign companies because of either incomplete
information or a too restricted client base10. The final sample includes 4211 products, corresponding to
55 separate tariffs (the units of our analysis) and issued by 16 companies, mainly part of bancassurance
groups.
About three-quarters of the tariffs in our sample (Figure 2) are targeted to retail underwriters, with a
minimum premium below € 100,000; only 4 tariffs have a minimum premium above € 1,000,000.
Figure 2: Characteristics of the unit-linked policies included in our sample: target market and distribution
(number of tariffs)
(a) distribution of tariffs by minimum premium
(b) distribution of tariffs by distribution channel
Note: (1) contracts that are distributed through different channels are counted more than once, therefore the total is higher than the number of
tariffs (7 tariffs are jointly distributed by banks and advisors and 1 by agents and banks).
Typically, a unit-linked policy gives the option to the underwriter to choose how to invest the premium
choosing among a list of predefined funds. Some policies also offer tariffs allowing to invest in predefined
portfolios of external funds12. Overall, in our sample there are 10,717 available investment options
(distinct combinations of policies and funds/portfolios) and 2,344 unique funds’ share classes13.
The synthetic index of costs, reduction in yield, of each unit-linked policy invested in a specific fund (or
in a fixed portfolio of funds) is compared to that of the corresponding mutual funds’ share classes sold
in the retail market14.
Twenty products are offered by a cross-border company specialized in private clients (over 1 million of minimum premium)
and limited distribution network in Italy; for 2 products the conditions of the contracts were not available for download on
the website.
11 Each product may have one or more tariffs (or contracts) connected to the same set of funds, or sharing the same product
structure and the same kind of life coverage. A tariff may be structured either to target certain clients with discounted charges
(e.g. minimum premium to access the contract, reinvesting from another contract from the same company or distributor, etc.)
or with a dedicated list of funds different from the rest of the product architecture.
12 Our sample includes 16 tariffs that offer a total of 113 fixed portfolios of funds. Moreover, there are 19 tariffs that offer
internally managed funds along with external ones. Internal funds are not considered in the analysis as they cannot be directly
acquired in the retail market.
13 Mutual funds may have different share classes depending on the targeted market segment. Unit-linked policies generally
invest in premium share classes or classes targeted to institutional investors, which have lower fees compared to classes
targeted to retail investors. For instance, let’s take the fund AA SGR fixed income investment grade included in two unit linked
products offered by two different companies. One company has a contract with the AA SGR which let it use the class I
reserved to institutional investors; the other has a contract which allow it to buy the (P) premium class. Retail investors have
access to the associated retail class (R): AA SGR fixed income investment grade R. However, all classes have the same underlying
asset allocation.
14 We excluded from the analysis 146 funds since we did not find the corresponding retail class: 101 funds have a retail class
valued in a different currency than the policy, other funds are not present in the Morningstar database.
Table 2: Fees and bonuses: descriptive statistics1
(in percentage points; observations in units)
Unit linked products
Policy fee
Loading costs
Annual management charge
OICR ongoing charges
Surrender charges2
Initial bonus3
Death bonus4
Coverage5
Mutual funds
Service entry fee
Entry charge6
Exit charge6
OICR ongoing charges
Std. Dev
1st quartile
Median
3rd quartile
Observ.
0.101
0.604
2.001
1.028
0.162
0.792
3.203
0.798
0.215
1.222
0.473
0.567
0.548
1.522
4.670
0.763
0.000
0.000
0.100
0.050
0.000
0.000
0.000
0.000
0.000
0.000
1.657
0.670
0.000
0.000
0.500
0.214
0.000
0.000
1.950
0.930
0.000
0.000
2.000
0.300
0.050
0.075
2.320
1.226
0.000
0.000
3.000
1.000
1.000
4.000
3.150
3.670
4.000
6.000
35.000
2.500
295,164
295,164
295,164
295,164
295,164
295,164
295,164
295,164
0.093
3.305
0.062
1.671
0.197
1.870
0.406
0.657
0.000
0.000
0.000
0.000
0.002
2.500
0.000
1.350
0.012
3.500
0.000
1.710
0.120
5.000
0.000
2.020
0.850
9.000
3.000
5.000
279,504
279,504
279,504
279,504
Source: own elaborations based on individual policy prospectuses and on Morningstar Direct.
Note: (1) Summary statistics for the conditions actually accessible in the market, removing the financial case in which the actuarial components
are not considered. For the overall sample used in the analysis see Table A.1 (2) Zeros for the great part of the sample as several tariffs have not
such a penalty and even when they have it, the long holding period (5, 10 20 years) make it not applicable. In order not to omit observation we
record this data as zero. (3) Initial bonus is present only in very few products. (4) Death bonus is present for every product; for higher ages (in our
sample, 70 years old) some contracts may set the bonus to zero. (5) Data are in tenth of percentage point. Some products set coverage charges at 0
when the age of the underwriter is high. (6) Depending on each intermediary, either entry charges or exit charges are applied. In the simulation we
assume that the highest of the two is applied.
Table 2 reports some descriptive statistics for each type of fees used in the analysis. The fees are calculated
for each investment option (tariff-fund combination) and for a given value of holding period, age of the
investor/underwriter and investment amount. From the table we see that unit-linked policies on average
have entry charges lower than mutual funds15. Policies’ surrender charges are often null as for long
holding periods they are set to zero. Ongoing charges for funds sold through policies are on average lower
than those applied for retail funds, as institutional (or premium) funds’ share classes (i.e. those acquired
by unit linked policies) have typically lower charges than retail ones. However, on an ongoing basis,
policies apply also annual management charges (2.0 percent on average) that are not applied by retail
funds.
Since exit charges are applied alternatively to entry charges, we assumed that the highest of the two applies.
4. Methodology
4.1 Measure of costs: Reduction in yield
In order to measure the cost of an insurance policy and a mutual fund we use the Reduction in Yield
(RIY), a synthetic measure of all fees and charges of a financial investment. Specifically, the RIY is defined
as the difference between the yield of a hypothetical investment gross of fees and the yield of the same
investment net of all fees and charges.
The cost differential between a unit-linked policy and a mutual fund is therefore given by:
?????????????,?,? = ??????_????,?,? ? ????_????,?,
????_????,? = ?????,? ? ??,?,
??????_????,?,? = ?????,? ? ??,?,?,
where, for a given year T, ?????,? is the gross yield (without any charge and fee) of an investment in a
specific mutual fund i, ??,? is the yield net of all fees and charges of an investment in the same fund i,
and ??,?,? is the yield net of all fees and charges of an investment in a unit-linked policy j that invests in
the same fund i. The details of the calculations are in Appendix A.2 Reduction in Yield step by step, where
actuarial details are also explicated.
4.2 Monte Carlo simulation
In order to compare RIYs of policies and funds we simulate daily return, both gross and net of fees, by
means of Monte Carlo simulation. We assume that the evolution of interest rates follows a process with
Gaussian innovation. We simulate 100 yield paths with a mean of 4 per cent and a standard deviation of
10. We used 3 levels of premium (10,000, 100,000 and 500,000 euros), in order to control for the
incidence of fixed costs and for the fact that policy discounts vary according to the amount invested. The
holding period has been set to 5, 10 and 20 years, that are typical for investment in insurance products.
We simulated the RIY for an investment both with and without the demographic component. To account
for the impact of the death bonus, which is a decreasing function of age, we set the age of the underwriter
to 40, 55 and 70 years.
Each simulated path is used to compare the RIYs of an investment in each policy-fund combination with
that of an investment in the corresponding mutual fund. We then assess the distributions of the simulated
cost differentials across funds, policies and companies (Section 5.1. ).
4.3 Multivariate analysis
We use a multivariate analysis to decompose the cost differential into its main components and to assess
how it depends on tariffs’ and funds’ characteristics (Sections
5.2. and 5.3. Breakdown of the cost).
In order to estimate the impact of the investment amount, the holding period, the age of the underwriter
and the distribution channel, we use the following OLS regression model:
(1) ??,?,?,? = ?0 + ?1 × ????,?,? + ?2 × ???????_???????,?,? + ?3 × ????,?,? +
??1
? ???
??1
+ ???? + ????? + ? ??? + ??,?,?,?
?=1
?=1
The dependent variable (??,?,?,? ) is the average16 cost differential of the 100 realizations of the simulation
between a policy and an investment fund for each tariff-fund combination (o) and a given value of the
investment/premium (i), holding period (h) and age (a)
On the right-hand side of the model, bank and ETF are dummy variables identifying policies distributed
through bank branches and policies invested in ETFs, while ??? and ?? are dummies respectively for fund
categories (c)17 and tariff fixed effects (p). Finally, ????,?,?, ???????_???????,?,? and ????,?,? are
categorical variables for the different predefined values of the amount invested, the holding period and
the investor’s age18.
In order to estimate the impact of each type of fee, we use the following model:
(2) ??,?,?,? = ?0 + ??
?=1 ?? × ??????_???????,o,i,h,a + ??=1 ?? × ????_???????,o,i,h,a
??1
+?1 × ????,?,? + ?2 × ???????_???????,?,? + ?3 × ????,?,? + ? ?? + ??,?,?,?
?=1
In this model we include two set of regressors: ??????_???????,o,i,h,a 19, which is the level of the fees
for a given policy fee category (p) and ????_???????,o,i,h,a 20, which is the level of fee for a given fund
fee category (f). The other regressors (????,?,?, ???????_???????,?,? and ????,?,? and ?? ) are defined as
in equation (1).
5. Results
In this section, we first describe the simulated distributions of the cost differentials across funds, policies
and companies; we then discuss the results of the multivariate analysis.
5.1. Simulated distributions of the cost differential
The average simulated cost differential varies greatly across insurers (Figure 3, panel a). In the “financial
case” (i.e. when we do not consider the actuarial component of the insurance contract and set to zero
both the coverage fee and the death bonus), the average cost differential varies from a minimum of 0,5
per cent to a maximum of almost 2 per cent. For each company, those with the lowest average differential
?100 (??????_???????
????? = ?????????????????
???????????????? =
? ????_???????
100 ?=1
17 Using the Morningstar global category, we get: allocation, alternative, commodities, convertibles, fixed income, equity,
commodity, and miscellaneous. We also add the category portfolio in order to account for those that use combination of external
funds
18 inv is set to 1, 2 and 3 when the investment amount is respectively equal to 10.000, 100.000 and 500.000 (3). Age: financial
case (0), 40 (1), 55, (2), 70 (3). Holding period: 5 (1), 10 (2), 20 (3).
19 Namely: Policy fee, Loading costs, Ongoing OICR policy, Surrender charge, Coverage charge, Bonus claim, Initial bonus.
20 Namely: Service entry fee, Entry charge fund, Ongoing OICR fund, Ongoing fees policy, Exit charge fund.
are mostly mutual insurance companies as they set on average lower annual management charges. The cost
differentials of domestic companies are not significantly different from those of foreign companies
operating under the Freedom to Provide Services regime.
There is high heterogeneity also across tariffs (Figure 3, panel b). The average cost differential ranges
from a minimum of 0.5 per cent to a maximum of 2.5 per cent. Within each tariff, the share of investment
options (tariff-fund combinations) with a positive differential is at least 87 per cent (for most tariffs there
is none to two combinations with non-positive differential). The interquartile range is also low, with a
sample average of around 0.8 per cent.
Figure 3: Distributions of the cost differential by undertaking and by tariff
(percentage)
(a) by undertaking
(b) by tariff
Note: For each company (panel a) and for each tariff (panel b), the chart shows the average, the median and the interquartile range of the
simulated cost differentials.
Figure 4 shows the distribution of the cost differential across all investment options when the policy does
not include a life insurance component (financial case) and when the policy includes a life insurance
component (actuarial case at three different ages). As in figure 3, the cost differential is almost always
positive. It tends to be higher in the actuarial case, reflecting the extra cost of the insurance coverage
which is only partially compensated by the death bonus. In the actuarial case, the cost differential tends
to increase with the age of the investor/underwriter, especially for longer holding periods (panel b and
c), since the coverage charge is a fixed percentage while the death bonus declines with age. In particular,
in panel c, the average differential is highest (1.8 per cent) for a 20-years contract underwritten by a 70year old investor.
Figure 5 shows that in the financial case the cost differential, as shown in Figure 4, increases with the
holding period (on the x-axis of each panel), as ongoing charges are on average higher for policies than
for funds21. Instead, in these bivariate figures, the cost differential varies only slightly with the investment
amount (across panels) and is compensated by longer holding periods.
The difference is mostly due to the fact that on top of “ongoing fund’s charges” unit linked policies also apply “annual
management charges”.
Figure 4: Distributions of the cost differential across tariffs and funds, by investor’s age and holding period:
financial case (without the life insurance component) versus the actuarial case (with the life insurance
component)
(percentage)
(a) 5 years of contract; 10,000 investment
(b) 10 years of contract; 10,000
investment
(c) 20 years of contract; 10,000
investment
Note: The charts show the mean, the median and the interquartile range of the simulated distribution of cost differentials across tariffs and funds for
a given value of investor’s age and a given value of the amount invested. “Financial” indicates the simulation without the actuarial component of the
insurance policy.
Figure 5: Distributions of the cost differential across tariffs and funds, by holding period and value of the
investment (“financial case”)
(percentages)
(a) investment 10,000
(b) investment 100,000
(c) investment 500,000
Note: The charts show the mean, the median and the interquartile range of the simulated distribution of cost differentials between policy tariffs and
funds for a given value of the holding period and a given value of the amount invested. Simulations do not include the actuarial component of the life
insurance (“financial case”).
5.2. Determinants of the cost differential
We estimated model (1) presented in section 4.3 across all tariffs-funds combinations and for each
predefined value of investment amount, age and holding period, both with and without tariff fixed effects.
Results are shown in table 3. Results are slightly different than those seen with the analysis above, because
of the multivariate context, in particular regarding the level of investment. As a robustness check we also
estimated the model applying tariffs’ investment constraints, that is excluding an observation whenever
the corresponding investment amount was below the minimum investment (above the maximum)
allowed by the tariff22, 23.
Considering the specification with tariff investment constraints and including tariff fixed effects (Table
3, column 4), we find that the cost differential decreases by 6 basis points on average when Investment
increases by one level24 and it increases by 20 and 6 basis points on average, respectively, when Age and
Holding period rise by one level25. The negative impact of Investment on the cost differential is mainly due
to the discounts offered by unit linked contracts to higher levels of investments26.
The positive impact of Age is explained by the cost of the actuarial component of unit linked policies,
and the decrease of the death bonus in age. Finally, the positive effect of Holding period is due to the fact
that ongoing charges are on average greater for policies than for mutual funds and the incidence of this
recurrent cost component increases with the holding period. The effects of Investment, Age and Holding
period are statistically significant and their sign is coherent across all model specifications.
Table 3: Determinants of the cost differential: impact of investment, age and holding period
Investment
Holding period
Constant
Lowest fixed
effect
R-square
Tariff constraint
Tariff FE
-0.0360***
(0.0016)
0.2060***
(0.0011)
0.0440***
(0.0016)
1.0882***
(0.0052)
-0.0412***
(0.0031)
0.2043***
(0.0015)
0.0576***
(0.0020)
1.1268***
(0.0092)
-0.0360***
(0.0013)
0.2084***
(0.0009)
0.0440***
(0.0013)
-0.0650***
(0.0025)
0.2081***
(0.0011)
0.0576***
(0.0016)
0.2164***
(0.0082)
0.2310***
(0.0109)
362,681
204,390
362,681
204,390
Notes: (1) standard errors, in parentheses; (2) Level of significance: (***) 0.01; (**) 0.05; (*) 0.1
We then estimate three augmented versions of model 1 to evaluate the contribution of funds’
characteristics (investment category, ETF and distribution channel). Results are shown in table 4. The
All contracts in our sample allow for values of age considered in the simulation.
This is not the default option as, for instance, in the Key Information Document (KID) companies calculate the RIY for a
premium of 10,000 euros regardless of the minimum investment required to invest in it.
24 From 10,000 to 100,000 euro or from 100,000 to 500,000 euro.
25 For age, from 40 to 55 years or from 55 to 70 years; for holding period, from 5 to 10 years or from 10 to 20 years.
26 The negative impact of Investment is higher in the model with tariff investment constraints. By applying the constraints, we
drop from the estimation sample the observations with a premium above the threshold set by the tariff, which are on average
more expensive than those that have not such a constraint.
cost differential is highest for money market funds (the reference category in the model; columns 1 and
2), as for low risk funds’ categories the cost of the retail funds’ share classes is generally lower than for
other funds’ categories (see Table A.3 for more details). The cost differential is on average between 11
and 15 basis points lower (column 2) for balanced, equity or fixed income funds than for money market
funds. Products investing in predefined funds’ portfolios have a slightly lower differential than that of
those investing in the balanced, equity and fixed income categories.
Table 4: Determinants of the cost differential: funds’ characteristics
Investment
Holding period
Balanced3
Alternative
Commodities
Convertibles
Equity
Fixed income
Miscellaneous
Portfolio4
-0.0362***
(0.0016)
0.2060***
(0.0011)
0.0435***
(0.0016)
-0.2044***
(0.0125)
-0.2989***
(0.0152)
-0.6268***
(0.0280)
-0.3080***
(0.0154)
-0.2448***
(0.0123)
-0.1904***
(0.0123)
0.3165***
(0.0301)
-0.1815***
(0.0159)
-0.0423***
(0.0031)
0.2043***
(0.0015)
0.0575***
(0.0021)
-0.1140***
(0.0143)
-0.1785***
(0.0180)
-0.3577***
(0.0330)
-0.2179***
(0.0184)
-0.1529***
(0.0140)
-0.1227***
(0.0139)
0.3534***
(0.0367)
-0.1866***
(0.0195)
-0.0360***
(0.0003)
0.2092***
(0.0006)
0.0440***
(0.0089)
-0.0648***
(0.0029)
0.2082***
(0.0013)
0.0576***
(0.0019)
-0.0362***
(0.0016)
0.1995***
(0.0011)
0.0456***
(0.0016)
-0.0462***
(0.0032)
0.1995***
(0.0015)
0.0599***
(0.0021)
0.5640***
(0.0238)
0.6367***
(0.0033)
0.2903***
(0.0046)
1.1290***
(0.0094)
204,390
Advisor
1.3094***
(0.0132)
1.2669***
(0.0165)
0.9051***
(0.0233)
0.9732***
(0.0085)
0.3864***
(0.0042)
1.0774***
(0.0052)
362,681
204,390
362,681
204,390
362,681
Constant
R-square
Tariff constraint
Tariff FE
Notes: (1) standard errors, in parentheses; (2) Level of significance: (***) 0.01; (**) 0.05; (*) 0.1; (3) Allocation in the global
Morningstar category, we used balanced as more widely used. (4) This category refers to the investment options with a predefined fixed
portfolio of funds, to which cannot be assigned any specific Morningstar global category.
Funds included in the predefined portfolios are usually cheaper since they are typically offered by
management companies that belong to the same financial group of the insurer as they can have an easier
access to institutional classes. The distribution of unit linked policies has a large impact on costs: the cost
differential is 64 basis points higher if it is sold by advisors rather than banks (column 4).
Finally, ETFs (listed index funds that we do not differentiate by class of investment) have on average a
cost differential 38 basis points higher than non-ETF funds, given that for ETFs’ retail classes have
significantly lower costs than other funds.
5.3. Breakdown of the cost differential
This section analyses the components of the simulated cost differential between unit-linked policies and
mutual funds using results from equation 2 presented in section 4.3 Multivariate analysis. The breakdown
is shown in Table 5 while the regression results are displayed in Table A.4. Columns refer to the same
specifications used in Table 3. The regressions control for investment amount, holding period, age and
tariffs fixed effects.
The overall cost differential, evaluated at the mean values of all cost components, ranges between 0.62
and 0.75 percentage points depending on model specification. When controlling for tariffs fixed effects
(columns 3 and 4) the differential reduces but remains sizable. Ongoing charges explain by far the biggest
part of the differential, with a positive impact ranging from 0.65 to 0.95 percentage points. As showed in
Table A.5, which includes all costs subcategories, they are the result of two components: annual management
fee, which is a remuneration for the insurance company management services27; ongoing charges of funds,
which remunerate the fund’s asset management company and are on average lower for policies than for
retail funds. Initial charges reduce the cost differential by around 10 basis points as they are generally
higher for retail funds than for policies. Finally, the contribution of exit charges is negligible.
Table 5: Breakdown of the simulated average cost differential1
(percentage points)
Cost component
Initial charge
Ongoing charges
Exit charges
Other charges
differential due to
charges
Constant
Fixed effect with
lowest value
Fixed effect with
highest value
controls at
minimum2
controls at
maximum3
min differential
max differential
-0.1238
0.9036
0.0098
-0.0445
0.7451
-0.0873
0.9462
0.0151
-0.0603
0.8137
-0.0795
0.6449
0.0054
0.0487
0.6194
-0.0168
0.6643
0.0106
0.0460
0.7041
0.2667
0.2630
0.0021
-0.0774
1.0853
1.1992
-0.1083
-0.2079
-0.0762
-0.1467
0.5653
0.5737
0.5962
0.6444
0.9035
1.5771
0.8688
1.6504
0.5453
2.3010
0.4800
2.5477
Note:(1) Table A.5 in the appendix give details on the individual fees (2) Premium is set to level 3, holding period to 1, age to 0. (3) Premium is
set to 1, holding period to 3 and age to 3.
It does not imply neither from the companies nor from the distributors an active management of the allocation across the
funds available. These services are sometimes sold within contracts only if the underwriter explicitly chooses them, paying an
additional fee.
6. Robustness checks
6.1 Controlling for rebates and switching costs
Our base simulation exercise did not consider rebates28 (since we do not have this information at the
contract level29) and switching costs (i.e. costs for changing the allocation of the portfolio). In order to
control for the robustness of our results to rebates and switching costs we use the model presented in
section 4.4, with no tariff investment constraints and no tariff fixed effects (as in table 5, column 1)30.
We hypothesize that rebates equal to 40 per cent of the funds’ ongoing charges and are applied on all unitlinked investment options31. Concerning switching costs, we hypothesize that investors rebalance their
portfolio 5 times during the holding period and that the corresponding additional costs are paid only
when they invest directly in mutual funds and not when they underwrite a unit linked policy as a certain
number of switch can be executed for free32. Under these upper bound hypotheses the simulated cost
differential decreases significantly but it remains positive and sizable. In particular, rebates lower ongoing
costs from 0.90 to 0.58 per cent and the overall simulated cost differential from 0.75 to 0.41 per cent (table
6, column 2). At the same time, switching costs increase the differential of funds’ initial charges (service
entry fee plus entry charge) from 0.12 to 0.74 percentage points and hence reduce the cost differential from
0.75 to 0.14 per cent (table 6, column 3).
Table 6: Impact of rebates and switching costs on the simulated cost differential1
Feature
Rebates
5 switches
Initial charge
Ongoing charges
Exit charges
Other charges
differential
-0.1238
0.9036
0.0098
-0.0445
0.7451
-0.1238
0.5835
0.0098
-0.0445
0.4136
-0.7427
0.9036
0.0172
-0.0497
0.1409
Constant
controls at minimum1
controls at maximum2
min differential
max differential
0.2706
-0.1083
0.5653
0.9035
1.5771
0.2706
-0.1083
0.5653
0.5720
1.2456
0.2706
-0.1083
0.5653
0.2993
0.9729
Note: (1) Based on a regression model without tariff investment constraints and tariff fixed effects. (1) Premium is set to level 3, holding period to
1, age to 0. (2) Premium is set to 1, holding period to 3 and age to 3.
Sometimes, the asset management company pays back to the insurance company a certain percentage of the fund’s ongoing
fees (rebates), which the law compels insurers to return to the underwriter. This is done through a discount in the annual
management fee.
29 Rebates are agreed between the insurer and the fund’s management company on a case by case level and applied to the
funds’ ongoing charges. The corresponding amount is then repaid by the insurer to the underwriter in the form of a reduction in
the annual management fee. In addition, they can change even on a monthly basis.
30 The specifications corresponding columns 2, 3 and 4 of Table 5 are in Table A.7, A.8, A.9.
31 On the basis of typical contracts, the 40 per cent discount assumed in the exercise can be considered as an upper bound
value; moreover, the exercise applies rebates to all tariffs and funds, although they are applied on a case by case basis.
32 This hypothesis should allow us to obtain an upper bound estimate of the impact of switching costs on the simulated cost
differential as it is quite a strict hypothesis that all the costs of exiting and entering (e.g. entry service fee and entry charge) a fund
are applied in each switch of mutual funds which is quite implausible. In addition, five portfolio switches represent a moderate
rebalancing rate for retail investors. Unit linked contracts typically allow the investor to make from 3 to 5 switches per year
for free and apply e a charge for additional switches.
6.2 Volume weighted differential
The analysis performed so far considered for each policy the distribution of cost differentials across all
funds offered by a tariff. We could not provide a synthetic measure of the cost differential based on the
funds actually chosen by the underwriter as we do not have contract-level information on the actual
amounts invested in each fund. In this paragraph, however, for each policy we compute an average cost
differential weighted by the share of each fund in the company overall unit linked portfolio, using data
from insurers’ supervisory reports33. We then compare this measure with an unweighted version, that is,
a simple average using uniform weights across all funds included in a product prospectus. Compared to
the unweighted benchmark, in the weighted formula funds may have lower (even a zero) or higher
weights. This exercise allows us to assess how the aggregate actual average cost differential incurred by
company customers compares to the cost differential distributions we discussed in section 5.
Panel a of Figure 6, shows that the two measures of the cost differential (weighted and unweighted) have
approximately the same distribution across policies: the difference in the average value is 0.03 per cent
and the median is basically the same. Panel b shows that the two measures have a different sign only in 5
cases (red dots, 1.3 per cent of occurrences). The differences between the weighted and the unweighted
averages lie close and around the 45-degrees line of the first quadrant (which identifies policies with the
same weighted and unweighted differential), showing that there is no clear bias. We can thus conclude
that using the actual holding by undertaking does not have a material impact on the estimation of the
cost differential between unit linked policies and investments funds.
Figure 6: Volume-weighted vs unweighted cost differential across policies
(percentage points)
a) distribution of the differential by policy
b) comparison of weighted and unweighted averages by
observation1
Source: weights are based on insurers’ supervisory reports on investments related to unit-linked policies; data are available only for Italian
undertakings.
Note: (1) The analysis is performed only on the financial case as it is the occurrence that could be susceptible of a change in the sign of the differential.
Figure 7 shows how the distribution across products of the two measures varies by company. In
particular, we observe that the weighted average cost differentials are close to the unweighted ones
without any clear pattern or bias.
This analysis is performed only for Italian undertakings as data for foreign companies are not available.
Figure 7: Distribution of the volume weighted and the unweighted cost differential across policies, by company
(percentage points)
Source: weights are based on insurers’ supervisory reports on investments related to unit-linked policies; data are available only for Italian
undertakings.
References
ESMA (2023), “Annual Report”, ESMA, June 2023, p. 22.
EIOPA (2023), “Costs and Past Performance Report”, EIOPA, December 2023, p. 18.
Greene, W.H. (2019), Econometric Analysis, Global Edition, Pearson.
Harpreet, A., Pardeep, K. (2017), “Performance Evaluation of Mutual Funds and Unit Linked Insurance
Plans in India: An Empirical Study of Equity-Based Funds”, International Journal of Emerging Trends in Science
and Technology, 4(8):5779-5798.
Morningstar (2021), “Global Investor Experience Study: Fees and Expenses”, Morningstar Report 2021,
p. 36.
Nadine, G., Carin, H., Hato, S. (2011), “On the Valuation of Investment Guarantees in Unit-Linked Life
Insurance: A Customer Perspective”, The Geneva Papers on Risk and Insurance. Issues and Practice, Vol. 36,
No. 1 (January 2011), pp. 3-29.
Pitacco, E., Olivieri, A. (2011), Introduction to Insurance Mathematics, Springer.
Signorini, L.F. (2023), Speech at ANIA Annual Assembly, IVASS, 4 July 2023.
Annexes
A.1 Costs
Unit-linked contracts
? Policy fee: fee paid by the underwriter for compensating administrative costs associated with the
issuance of the policy. The fee is subtracted upfront from the investment amount.
? Loading costs: percentage of the premium paid to cover distribution and other residual
administrative costs. Sometimes it includes the death benefit coverage costs.
? Annual management charges: remuneration of the insurance company for the maintenance of
the funds’ list and for the eventual extraordinary liquidation of the funds. Often it includes the
death benefit coverage costs.
? Ongoing fund’s charges: remuneration of the fund’s management company. Insurance
companies have access to premium or institutional funds’ classes with lower fees than those
applied on retail mutual funds.
? Death benefit coverage costs: insurance premium paid for remunerating the bonus provided
in case of death of the underwriter. It may be included among loading costs, annual management
charges or as additional recurring costs.
? Surrender charge: fee paid for liquidating the investment. It decreases with the holding period
of the policy.
? Coverage fee: either charged up-front or ongoing, it is needed to cover the actuarial costs of the
death bonus and remunerate that service.
? Initial bonus: bonus given as increase in the quotes underwritten
? Death bonus: increase in the quotes at the death of the insured person, it is the compulsory
insurance content of the unit-linked contract.
Mutual funds
Service entry/exit fee: remuneration of third parties for acquiring or liquidating the position.
Entry charge: remuneration of the distributor who can apply a percentage (from 0 to 100 per
cent) of the value stated in the Key investor information document (KIID); the percentage is
agreed on by the management company and the distributor. It is applied alternatively to the exit
charge.
Exit charge: remuneration of the distributor who can apply a percentage (from 0 to 100 per
cent) of the value stated in the Key investor information document (KIID); the percentage is
agreed on by the management company and the distributor. It is applied alternatively to the entry
charge. In the simulation we assume that the distributor applies it only when it is greater than
entry charge.
Ongoing fund’s charges: remuneration of the fund’s management company paid by the
investor in retail mutual funds. It is a comprehensive measure of ongoing costs, including
management fee, transaction costs and all other recurring fees.
A.2 Reduction in Yield step by step
Let be ????? ? the gross yield (without any charge) in year T, ?? the yield considering all fees and charges
of the investment in the fund, and ?? the yield considering all fees and charges of the investment in the
same fund but supplied through the policy, then the two RIYs are:
??????_???? = ?????? ? ?? and ????_???? = ?????? ? ??
Let ?? and ?? be the values of insurance and mutual funds respectively after the costs that
insured/investor should pay34 according to the respective contracts, but before the surrender charge and
exit cost. So at fixed time ?, the net yield ?? and ?? are defined as:
?? =
? ?(??????????? ??????) ?
( ? ??????? ?????????? )
? ? for policy insurance
?? =
? ?(?????? ??????) ?
( ?
??????? ??????????
? ? for mutual fund
The policy yield depends on the time ?, moment of the death of the underwriter, bonus ??, depending
on age ? of the insured, triggered by the event and for a generic time ?, we calculate it as follows:
?? ? (? + ??+? )
?? = (
) ??
??????? ??????????
For policies we have also the probability of deceasing or surviving is integrated as follows: Since the time
? is not known at beginning of the contract, we calculate a RIY that is weighted probability
??????_???? = ???1
?=0 ??? (?????? ? ?? )+ ??? (????? ? ? ?? )
????_???? = ???1
?=0 ??? (?????? ? ?? )+ ??? (????? ? ? ?? )
Where ??? is the probability to death between age ? + ? and ? + ? + 1, for an insurers of age ? at
time 0, while ??? is the survivor probability at age ? + ?; obviously ???1
?=0 ??? + ? ?? = 1. For our
analysis we use the mortality table sim2002 published by ISTAT. For mutual funds we work under the
assumption that heirs close the positions when the investor has died, paying the exit cost if present.
Cfr. Section 1 or Annex A.1 Costs
A.3 Figures and Tables
Table A.1: Fees and bonuses: descriptive statistics with the financial case
(in percentage points; observations in units)
Unit linked products
Policy fee
Loading costs
Annual management charge
OICR ongoing charges
Surrender charges1
Initial bonus2
Death bonus3
Coverage4
Mutual funds
Service entry fee
Entry charge5
Exit charge5
OICR ongoing charges
Std. Dev
1st quartile
Median
3rd quartile
Observ.
0.101
0.604
2.001
1.028
0.162
0.792
3.703
0.598
0.215
1.222
0.473
0.567
0.548
1.522
5.325
0.746
0.000
0.000
0.100
0.050
0.000
0.000
0.000
0.000
0.000
0.000
1.657
0.670
0.000
0.000
1.000
0.000
0.000
0.000
1.950
0.930
0.000
0.000
2.500
0.227
0.050
0.075
2.320
1.226
0.000
0.000
5.000
1.000
1.000
4.000
3.150
3.670
4.000
6.000
35.000
2.500
393,552
393,552
393,552
393,552
393,552
393,552
393,552
393,552
0.093
3.305
0.062
1.671
0.197
1.870
0.406
0.657
0.000
0.000
0.000
0.000
0.002
2.500
0.000
1.350
0.012
3.500
0.000
1.710
0.120
5.000
0.000
2.020
0.850
9.000
3.000
5.000
372,672
372,672
372,672
372,672
Source: own elaborations based on individual policy prospectuses and on Morningstar Direct.
Note: (1) Zeros for the great part of the sample as several tariffs have not such a penalty and even when they have it, the long
holding period (5, 10 20 years) make it not applicable. In order not to omit observation we record this data as zero. (2) Initial
bonus is present only in very few products. (3) Death bonus is present for every product; for higher ages (in our sample, 70
years old) some contracts may set the bonus to zero. We set the death bonus at 0 when we do not consider the insurance
coverage (“financial case”). (4) Data are in tenth of percentage point. Some products set coverage charges at 0 when the age
of the underwriter is high; we set them at 0 when we do not consider the insurance coverage (the “financial” case). (5)
Depending on each intermediary, either entry charges or exit charges are applied. In the simulation we assume that the highest
of the two is applied.
Table A.2: asset classes by global Morningstar category
(in percentage points; observations in units)
Global category
Balanced1
Alternative
Commodities
Convertibles
Equity
Fixed Income
Miscellaneous
Money Market
Portfolio2
Total
Number of underlying
investment options
10717
percentage
14.52
45.53
32.79
Note: (1) Allocation in the global Morningstar category, we used balanced as more widely used. (2) Not in Morningstar global category, but
reflecting the products that supply external funds in fixed combinations.
Table A.3: Ongoing charges by category of funds
(percentage point)
Morningstar global fund
category
Annual
management fee
Allocation1
Alternative
Commodities
Convertibles
Equity
Fixed Income
Miscellaneous
Money Market
Portfolio2
Total
Ongoing charges of
funds sold by policies
Ongoing charges of
retail funds
Difference of funds’
ongoing charges
(a)-(b)
-0.63
-0.77
-0.79
-0.61
-0.73
-0.50
-0 .53
-0.10
-0.96
-0.64
Note: (1) Allocation in the global Morningstar category, we used balanced as more widely used. (2) Not in Morningstar global category, but
reflecting the products that supply external funds in fixed combinations.
Table A.4: Regressions used to evaluate the impact of the different types of cost on the simulated cost differential 1
Policy fee
Service entry fee
Loading costs
Entry charge fund
Ongoing OICR fund
Ongoing OICR policy
Ongoing fees policy
Surrender charge
Exit charge fund
Coverage charge3
Bonus claim
Initial bonus
Holding period
Investment
Constant
R-square
Tariff constraint
Tariff FE
0.1566***
(0.0040)
-0.5589***
(0.0062)
0.0406***
(0.0007)
-0.0340***
(0.0005)
-0.9905***
(0.0017)
0.9199***
(0.0018)
0.8061***
(0.0022)
0.2778***
(0.0012)
0.0040***
(0.0001)
0.0438***
(0.0013)
-0.0025
(0.0019)
-0.0749***
(0.0005)
0.0741***
(0.0011)
0.1346***
(0.0007)
-0.0608***
(0.0013)
0.2667***
(0.0072)
0.6806
367,167
0.5433***
(0.0371)
-0.5783***
(0.0067)
0.0355***
(0.0009)
-0.0333***
(0.0006)
-0.9882***
(0.0020)
0.9562***
(0.0023)
0.8068***
(0.0029)
0.3368***
(0.0017)
-0.0011***
(0.0002)
0.0678***
(0.0015)
-0.0059***
(0.0022)
-0.0710***
(0.0006)
0.0996***
(0.0013)
0.1258***
(0.0009)
-0.1025***
(0.0021)
0.2630***
(0.0094)
0.7309
207,168
0.1135***
(0.0042)
-0.3418***
(0.0058)
0.0611***
(0.0024)
-0.0291***
(0.0005)
-0.9511***
(0.0016)
0.8920***
(0.0020)
0.6582***
(0.0044)
0.2092***
(0.0018)
0.0104***
(0.0001)
0.0240***
(0.0013)
0.0081*
(0.0018)
0.0128***
(0.0016)
0.0657***
(0.0010)
0.1488***
(0.0007)
-0.0473***
(0.0013)
0.0021
(0.0115)
0.7331
367,167
0.4084***
(0.0470)
-0.3006***
(0.0058)
0.1010
(0.0030)
-0.0276***
(0.0005)
-0.9396***
(0.0019)
0.9337***
(0.0024)
0.6369***
(0.0049)
0.2036***
(0.0024)
0.0092***
(0.0002)
0.0475***
(0.0015)
0.0097*
(0.0020)
0.0150***
(0.0017)
0.0891***
(0.0011)
0.1519***
(0.0008)
-0.0786***
(0.0021)
-0.0774***
(0.0140)
0.7958
207,168
Notes: (1) coefficients should be interpreted as how much of the costs is translated to the differential, for example 4 per cent of Loading costs are
translated to the differential (2) standard errors, in parenthesis; (3) p-value are calculated as percentage of simulations with a coefficient of opposite
sign to the one displayed in the table out of 100 simulations Level of significance: (***) 0.01; (**) 0.05; (*) 0.1; (3) data in thousandths, not in
percentages as the other coefficients
Table A.5: Breakdown of the simulated cost differential1
(percentage points)
Cost component
policy fee
service entry fee
loading costs
entry charge funds
Initial charge
ongoing funds’
charges (policy)
ongoing funds’
charges (retail fund)
annual management
fee (policy)
Ongoing charges
surrender policy
exit charge2
Exit charges
coverage charge
death bonus
initial bonus
Other charges
differential due to
charges
constant
fixed effect with
lowest value
fixed effect with
highest value
controls at
minimum2
controls at
maximum3
min differential
max differential
0.0158
-0.0518
0.0245
-0.1124
-0.1238
0.9457
0.0549
-0.0536
0.0215
-0.1101
-0.0873
0.9830
0.0115
-0.0317
0.0369
-0.0962
-0.0795
0.9170
0.0413
-0.0279
0.0611
-0.0912
-0.0168
0.9599
-1.6549
-1.6511
-1.5891
-1.5699
1.6129
1.6143
1.3169
1.2743
0.9036
0.0071
0.0027
0.0098
0.0166
0.0148
-0.0594
-0.0445
0.7451
0.9462
0.0110
0.0042
0.0151
0.0202
-0.0041
-0.0563
-0.0603
0.8137
0.6449
0.0039
0.0015
0.0054
0.0125
0.0385
0.0101
0.0487
0.6194
0.6643
0.0077
0.0029
0.0106
0.0122
0.0341
0.0119
0.0460
0.7041
0.2667
0.2630
0.0021
-0.0774
1.0853
1.1992
-0.1083
-0.2079
-0.0762
-0.1467
0.5653
0.5737
0.5962
0.6444
0.9035
1.5771
0.8688
1.6504
0.5453
2.3010
0.4800
2.5477
Note: (1) Policies’ charges have a positive coefficient because they increase the cost differential, while mutual funds’ charges have a negative coefficient
because they decrease the differential (2) It is remarkable, however, that funds’ exit charges have a positive coefficient rather than a negative one. The
impact of exit charges for those funds where we applied them (when greater than entry charges) is lower than the one of entry charges on other funds,
as a consequence the differential is higher for those applying exit charges and therefore the sign is positive rather than negative (3) premium is at level
3, holding period at 1, age at 0. (3) premium is at level 1, holding period at 3 and age at 3.
Table A.6: Impact of rebates and switching costs on the simulated cost differential (first model)
Feature
Equity
Rebates
5 switch
0.0203
-0.0499
0.0036
0.0000
-0.0260
0.2234
Money
Market
0.0158
-0.0518
0.0245
-0.1124
-0.1238
0.9457
specification
Policy fee
Service entry fee
Loading costs
Entry charge funds
Initial charge
Ongoing policy
Ongoing fund
Annual
management fee
Ongoing charges
surrender policy
exit charge
Exit charges
Coverage charge
Death bonus
Initial bonus
Other charges
differential
0.0158
-0.0518
0.0245
-0.1124
-0.1238
0.9457
0.0158
-0.0499
0.0234
-0.0355
-0.0461
0.1903
0.0158
-0.0499
0.0234
-0.1149
-0.1256
1.0950
0.0158
-0.2495
0.0245
-0.5324
-0.7416
0.9457
-1.6549
-0.2354
-1.6549
-0.3088
-1.9057
-1.6561
1.6129
1.6757
1.2814
1.2640
1.7355
1.6254
0.9036
0.0071
0.0027
0.0098
0.0166
0.0148
-0.0594
-0.0445
0.7451
1.6636
0.0057
0.0000
0.0057
0.0166
0.0095
-0.0441
-0.0346
1.6087
0.5722
0.0071
0.0027
0.0098
0.0166
0.0148
-0.0594
-0.0445
0.4136
1.1455
0.0057
0.0000
0.0057
0.0166
0.0096
-0.0594
-0.0497
1.0552
0.9249
0.0057
0.0029
0.0086
0.0166
0.0096
-0.0594
-0.0497
0.7581
0.9150
0.0071
0.0101
0.0172
0.0166
0.0096
-0.0594
-0.0497
0.1409
Constant
controls at
minimum1
controls at
maximum2
min differential
max differential
0.2667
-0.1083
0.2667
-0.1083
0.2667
-0.1083
0.2667
-0.1083
0.2667
-0.1083
0.2667
-0.1083
0.5653
0.5653
0.5653
0.5653
0.5653
0.5653
0.9035
1.5771
1.7671
2.4407
0.5720
1.2456
1.2136
1.8872
0.9165
1.5901
0.2993
0.9729
Note: (1) premium is at level 3, holding period at 1, age at 0. (2) premium is at level 1, holding period at 3 and age at 3.
Table A.7: Impact of rebates and switching costs on the simulated cost differential (second model)
Feature
Equity
Rebates
5 switch
0.0703
-0.0516
0.0032
0.0000
0.0218
0.2322
Money
Market
0.0549
-0.0516
0.0205
-0.0347
-0.0110
0.1978
specification
Policy fee
Service entry fee
Loading costs
Entry charge funds
Initial charge
Ongoing policy
Ongoing fund
Annual
management fee
Ongoing charges
surrender policy
exit charge
Exit charges
Coverage charge
Death bonus
Initial bonus
Other charges
differential
0.0549
-0.0536
0.0215
-0.1101
-0.0873
0.9830
0.0549
-0.0516
0.0205
-0.1125
-0.0888
1.1382
0.0549
-0.0536
0.0215
-0.1101
-0.0873
0.9830
0.0549
-0.2582
0.0215
-0.5214
-0.7033
0.9830
-1.6511
-0.2348
-0.3081
-1.9013
-1.6511
-1.6522
1.6143
1.6771
1.2651
1.7370
1.2825
1.6268
0.9462
0.0110
0.0042
0.0151
0.0202
-0.0041
-0.0563
-0.0603
0.8137
1.6744
0.0088
0.0000
0.0088
0.0202
-0.0026
-0.0418
-0.0444
1.6606
1.1548
0.0088
0.0000
0.0088
0.0202
-0.0026
-0.0563
-0.0589
1.0936
0.9740
0.0088
0.0045
0.0133
0.0202
-0.0026
-0.0563
-0.0589
0.8396
0.6144
0.0110
0.0042
0.0151
0.0202
-0.0041
-0.0563
-0.0603
0.4819
0.9576
0.0110
0.0157
0.0266
0.0202
-0.0026
-0.0563
-0.0589
0.2220
Constant
Fixed effect with
highest value
controls at
minimum1
controls at
maximum2
min differential
max differential
0.2630
0.2630
0.2630
0.2630
0.2630
0.2630
-0.2079
-0.2079
-0.2079
-0.2079
-0.2079
-0.2079
0.5737
0.5737
0.5737
0.5737
0.5737
0.5737
0.8688
1.6504
1.7157
2.4973
1.1487
1.9303
0.8947
1.6763
0.5370
1.3186
0.2771
1.0587
Note: (1) premium is at level 3, holding period at 1, age at 0. (2) premium is at level 1, holding period at 3 and age at 3.
Table A.8: Impact of rebates and switching costs on the simulated cost differential (third model)
Feature
Equity
Rebates
5 switch
0.0147
-0.0305
0.0055
0.0000
-0.0104
0.2166
Money
Market
0.0115
-0.0305
0.0352
-0.0911
-0.0750
0.9170
specification
Policy fee
Service entry fee
Loading costs
Entry charge funds
Initial charge
Ongoing policy
Ongoing fund
Annual
management fee
Ongoing charges
surrender policy
exit charge
Exit charges
Coverage charge
Death bonus
Initial bonus
Other charges
differential
0.0115
-0.0305
0.0352
-0.0911
-0.0750
0.9170
0.0115
-0.0305
0.0352
-0.0303
-0.0142
0.1845
0.0115
-0.0305
0.0352
-0.0983
-0.0822
1.0618
0.0115
-0.1526
0.0352
-0.4557
-0.5616
0.9170
-1.5902
-0.2260
-1.5902
-0.2965
-1.8299
-1.5902
1.3272
1.3682
1.0565
1.0321
1.4171
1.3272
0.6540
0.0039
0.0014
0.0053
0.0125
0.0250
0.0101
0.0351
0.6194
1.3588
0.0031
0.0000
0.0031
0.0125
0.0247
0.0075
0.0322
1.3838
0.3833
0.0039
0.0014
0.0053
0.0125
0.0250
0.0101
0.0351
0.3487
0.9201
0.0031
0.0000
0.0031
0.0125
0.0250
0.0101
0.0351
0.9441
0.6490
0.0031
0.0016
0.0047
0.0125
0.0250
0.0101
0.0351
0.6067
0.6540
0.0039
0.0055
0.0094
0.0125
0.0250
0.0101
0.0351
0.1369
Fixed effect with
lowest value
Fixed effect with
highest value
controls at
minimum1
controls at
maximum2
min differential
max differential
0.0021
0.0021
0.0021
0.0021
0.0021
0.0021
0.9759
0.9759
0.9759
0.9759
0.9759
0.9759
-0.0762
-0.0762
-0.0762
-0.0762
-0.0762
-0.0762
0.5962
0.5962
0.5962
0.5962
0.5962
0.5962
0.5453
1.2177
1.3097
1.9821
0.2747
0.9471
0.8700
1.5424
0.5326
1.2050
0.0646
0.7370
Note: (1) premium is at level 3, holding period at 1, age at 0. (2) premium is at level 1, holding period at 3 and age at 3.
Table A.9: Impact of rebates and switching costs on the simulated cost differential (fourth model)
Feature
specification
Policy fee
Service entry fee
Loading costs
Entry charge funds
Initial charge
Ongoing policy
Ongoing fund
Annual
management fee
Ongoing charges
surrender policy
exit charge
Exit charges
Coverage charge
Death bonus
Initial bonus
Other charges
differential
Constant1
Fixed effect with
highest value
controls at
minimum1
controls at
maximum2
min differential
max differential
Equity
Rebates
5 switch
Money
Market
0.0413
-0.0279
0.0611
-0.0912
-0.0168
0.0528
-0.0268
0.0091
0.0000
0.0350
0.0413
-0.0279
0.0611
-0.0912
-0.0168
0.0413
-0.0268
0.0582
-0.0288
0.0438
0.0413
-0.0268
0.0582
-0.0932
-0.0206
0.0413
-0.1342
0.0611
-0.4322
-0.4641
0.9599
0.2267
0.9599
0.1931
1.1115
0.9599
-1.5699
-0.2233
-1.5699
-0.2930
-1.8078
-1.5710
1.2743
1.3239
1.0124
0.9987
1.3712
1.2842
0.6643
0.0077
0.0029
0.0106
0.0122
0.0341
0.0119
0.0460
0.7041
1.3274
0.0061
0.0000
0.0061
0.0122
0.0219
0.0088
0.0307
1.3992
0.4024
0.0077
0.0029
0.0106
0.0122
0.0341
0.0119
0.0460
0.4422
0.8989
0.0061
0.0000
0.0061
0.0122
0.0221
0.0119
0.0340
0.9828
0.6749
0.0061
0.0032
0.0093
0.0122
0.0221
0.0119
0.0340
0.6976
0.6731
0.0077
0.0110
0.0187
0.0122
0.0221
0.0119
0.0340
0.2617
-0.0774
-0.0774
-0.0774
-0.0774
-0.0774
-0.0774
1.1600
1.1600
1.1600
1.1600
1.1600
1.1600
-0.1467
-0.1467
-0.1467
-0.1467
-0.1467
-0.1467
0.6444
0.6444
0.6444
0.6444
0.6444
0.6444
0.4800
1.2711
1.1751
1.9662
0.2181
1.0092
0.7587
1.5498
0.4735
1.2646
0.0376
0.8287
Note: (1) premium is at level 3, holding period at 1, age at 0. (2) premium is at level 1, holding period at 3 and age at 3.