Response to joint Discussion Paper on Key Information Documents (KIDs)
Go back
Market risk could be measured using Value-at-Risk provided that the bulk of the risk is not in the tail, in which case TailVaR should be used. The choice of the parameters (e.g. confidence interval) are at least as important as the choice of the measure. In addition to this quantitative element, we would add as a qualitative measure the disclosure of product design features that have proven problematic in the past, based on the criteria set out in Recital 18 and through the comprehension alert.
Regarding credit risk, we would favour credit value at risk over CDS spreads given that the latter is a traded instrument whose price can diverge temporarily from its fundamental value due to liquidity constraints, market irrational behaviour or excessive risk appetite.
Regarding liquidity risk we would propose using the historical volatility of the bid-offer spread over a meaningful period. While a static picture of a bid-offer spread does not provide much information, its fluctuations over a sample period long enough to include times of stress provides much more interesting information about potential liquidity declines in times of stress.
We agree with the problems identified in the discussion paper with using the distribution of historical returns.
The latter approach might be more resource intensive for the supervisors and be gamed. We therefore reject the second suggestion and strongly favour the first one.
Typically structured products including a principal protection only at maturity face a higher risk of loss in case of early redemption whereas other products not including this feature have a risk profile more proportional to their maturity.
Alternatively and consistent with our response to Question 13, the time frame used for the performance scenarios could be the life of the product, provided that the presentation of the scenarios displays clearly the evolution of the pay-out over time and linked to the holding period.
In addition during the debate at Level 1, several proposals were made to include the taxation impact in the KID (amongst others by Parliament rapporteur Pervenche Berès) but all of these failed, and in the end the co-legislators agreed to only include a brief statement that national tax regimes may impact the actual pay-out (Article 8.3.(d)(v)).
As the co-legislators agreed it was not appropriate for manufacturers or distributors to look into the tax impact of every individual PRIIP on their clients, we believe that tax information should not be considered relevant for this particular section. Instead, inserting the word “gross” (“expected gross returns”) should warn consumers that these are pre-tax returns (in line with the Level 1 drafting referred above), which nevertheless can be compared easily amongst each other based on fictional gross amounts.
As risk can often be shifted in the tail of the distribution in structured products, it is of utmost importance that the scenarios capture it and therefore that the range chosen be wide enough. In addition, any market trigger or other event affecting the payoff during the life of the product should appear in the scenarios.
In any case, we recommend taking an uneven number of scenarios to have an equal number of upside and downside scenarios, in addition to the neutral scenario. In our view, not properly balancing the scenarios would be in violation of the Level 1 text, which states in Recital 13: “To meet the needs of retail investors, it is necessary to ensure that information on PRIIPs is accurate, fair, clear and not misleading for those retail investors.”
While it may give a misleading impression of simplicity, we are not convinced that providing a multidimensional indicator detailing each type of risk would help most retail investors to make a more informed decision. Consumer testing performed by independent non-industry experts should be used to validate this point.
Regarding the different examples provided, we like the Belgian and Dutch indicators shown in the discussion paper, but do not have any strong preference as long as the indicator includes at least five risk categories. Consumer testing should be used to determine which presentation enables investors to choose the most appropriate product.
We do not like however the indication typically lower rewards / typically higher rewards" in the UCITS KII synthetic indicator as it appeals to retail investors' greed from a behavioural perspective, which is problematic especially in the current context of very low interest rates."
As a second option the example presenting probability in a number of frequencies is also clear and provides a more detailed picture at one point in time, even though it does not display the impact of the holding period. In turn this makes the question of the time frame used for the performance scenarios more acute.
Again we believe that the ultimate choice should be based on consumer testing.
Consistent with our earlier comments, we favour 2B but believe that consumer testing should determine the final choice.
Finance Watch proposed some clarifications and examples in our policy paper Product rules for packaged retail products: why, when, how? " (pages 16-17).
[The PRIIP] invests in underlying assets that are not commonly invested in by retail investors:
We support the proposal to take inspiration from UCITS Article 50.
[The PRIIP] uses a number of different mechanisms to calculate the final return of the investment, creating a greater risk of misunderstanding on the part of the retail investor
We support using ESMA's opinion on types of complex products and tailoring it to fit the specific context of the PRIIPs Regulation.
The investment’s pay-off takes advantage of retail investor’s behavioural biases, such as a teaser rate followed by a much higher floating conditional rate, or an iterative formula
We support further work by the ESAs to develop a common understanding and identify the mechanisms taking advantage of such biases.
As discussed in our policy paper , a number of such mechanisms have already been identified by national supervisors such as the UK's FCA and could be used as a basis (e.g. teaser rate, products designed to frame consumer choice in a potentially misleading way, complex cost structures and claim procedures, “gimmick” packaging features, use of product names implying greater safety than possible etc.).
Additionally, as the Joint Committee recognizes, the Level 1 text lists three criteria that “in particular” should be taken into account to qualify a product as not simple and difficult (for retail investors) to understand. In fact, the original list of criteria as proposed by the Parliament listed 6 features, with the alert being displayed “if one or more” of the criteria would be met. The list was reduced to 4 features in the report as adopted by ECON and the Parliament plenary, and to 3 features during the trialogue phase. However, during that process it was understood that the 3-4 features would be indicative and hence the wording “especially if” was agreed during the final trialogue on 31 March 2014. During legal-linguistic verification, this word was changed to “in particular”, and hence we believe that the list should indeed be interpreted as non-exhaustive.
The above means that national supervisors will have discretion to extend the list of criteria that require the publication of the comprehension alert. As we share the ESAs' concern that this may lead to a lack of harmonised application of the comprehension alert in different member states, we recommend considering further (Level 3) guidance from the ESAs in this area."
Consumers benefit from uniform simple messages about the objectives of an investment product, and we do not agree that it is difficult for manufacturers to explicitly disclose the objective of a PRIIP.
Typically such a sentence could be as follows: “an investor purchasing this investment product believes that EU telecom companies stocks will rise over the next five years and NEVER decline by more than 40% at any time during that period.”
The reason for this is that there is a huge difference between understanding how a financial product works and being able to assess the risks attached or understanding the implicit market view taken.
This would therefore contribute greatly in our view to retail investors making informed decisions, as it would highlight clearly the view that they are taking and might make them think again if they do not share this view.
2: Do you agree with the description of the consumer´s perspective on risk expressed in the Key Questions?
We do agree with the description of the consumer's perspective on risk expressed in the Key Questions.3: Do your agree that market, credit and liquidity risk are the main risks for PRIIPs? Do you agree with the definitions the ESA’s propose for these?
We agree with both.4: Do you have a view on the most appropriate measure(s) or combinations of these to be used to evaluate each type of risk? Do you consider some risk measures not appropriate in the PRIIPs context? Why? Please take into account access to data.
As a general comment, we are not proponents of the Value-at-Risk metric due to its well-known weaknesses - namely assuming statistically independent events and underestimating rare events. However, in the specific context of the summary indicator for PRIIPs, we recognise that this metric is one option enabling the integration of risks and providing a ballpark estimate.Market risk could be measured using Value-at-Risk provided that the bulk of the risk is not in the tail, in which case TailVaR should be used. The choice of the parameters (e.g. confidence interval) are at least as important as the choice of the measure. In addition to this quantitative element, we would add as a qualitative measure the disclosure of product design features that have proven problematic in the past, based on the criteria set out in Recital 18 and through the comprehension alert.
Regarding credit risk, we would favour credit value at risk over CDS spreads given that the latter is a traded instrument whose price can diverge temporarily from its fundamental value due to liquidity constraints, market irrational behaviour or excessive risk appetite.
Regarding liquidity risk we would propose using the historical volatility of the bid-offer spread over a meaningful period. While a static picture of a bid-offer spread does not provide much information, its fluctuations over a sample period long enough to include times of stress provides much more interesting information about potential liquidity declines in times of stress.
6: Do you think that performance scenarios should include or be based on probabilistic modelling, or instead show possible outcomes relevant for the payouts feasible under the PRIIP but without any implications as to their likelihood?
We think that performance scenarios should be based on probabilistic modelling using forward looking simulations.We agree with the problems identified in the discussion paper with using the distribution of historical returns.
7: How would you ensure a consistent approach across both firms and products were a modelling approach to be adopted?
A standard model could be defined based on the ones currently used by financial institutions that are fairly similar to one another. Alternatively each institution's model could be benchmarked against a standard portfolio of investment products.The latter approach might be more resource intensive for the supervisors and be gamed. We therefore reject the second suggestion and strongly favour the first one.
8: What time frames do you think would be appropriate for the performance scenarios?
We believe that the time frame used for the performance scenarios could be the time frame that gives rise to the highest potential loss, to be chosen between the life of the product, the recommended holding period or any other period.Typically structured products including a principal protection only at maturity face a higher risk of loss in case of early redemption whereas other products not including this feature have a risk profile more proportional to their maturity.
Alternatively and consistent with our response to Question 13, the time frame used for the performance scenarios could be the life of the product, provided that the presentation of the scenarios displays clearly the evolution of the pay-out over time and linked to the holding period.
9: Do you think that performance scenarios should include absolute figures, monetary amounts or percentages or a combination of these?
We tend to believe that performance scenarios should include a standard monetary amount of 1000 euros, but consumer testing should be used to determine which option is best.10: Are you aware of any practical issues that might arise with performance scenarios presented net of costs?
We are not aware of any unresolvable practical issue that might arise with performance scenarios presented net of costs (including embedded costs) and believe on the contrary that it is indispensable for the scenarios to be meaningful.In addition during the debate at Level 1, several proposals were made to include the taxation impact in the KID (amongst others by Parliament rapporteur Pervenche Berès) but all of these failed, and in the end the co-legislators agreed to only include a brief statement that national tax regimes may impact the actual pay-out (Article 8.3.(d)(v)).
As the co-legislators agreed it was not appropriate for manufacturers or distributors to look into the tax impact of every individual PRIIP on their clients, we believe that tax information should not be considered relevant for this particular section. Instead, inserting the word “gross” (“expected gross returns”) should warn consumers that these are pre-tax returns (in line with the Level 1 drafting referred above), which nevertheless can be compared easily amongst each other based on fictional gross amounts.
11: Do you have any preferences in terms of the number or range of scenarios presented? Please explain.
We support the choice of three scenarios based on their probability of occurrence.As risk can often be shifted in the tail of the distribution in structured products, it is of utmost importance that the scenarios capture it and therefore that the range chosen be wide enough. In addition, any market trigger or other event affecting the payoff during the life of the product should appear in the scenarios.
In any case, we recommend taking an uneven number of scenarios to have an equal number of upside and downside scenarios, in addition to the neutral scenario. In our view, not properly balancing the scenarios would be in violation of the Level 1 text, which states in Recital 13: “To meet the needs of retail investors, it is necessary to ensure that information on PRIIPs is accurate, fair, clear and not misleading for those retail investors.”
12: Do you have any views, positive or negative, on the different examples for presentation of a summary risk indicator? Please outline advantages and disadvantages, and provide any other examples that you are aware of that you think would be useful.
A single visual element would be clearly more appealing to retail investors.While it may give a misleading impression of simplicity, we are not convinced that providing a multidimensional indicator detailing each type of risk would help most retail investors to make a more informed decision. Consumer testing performed by independent non-industry experts should be used to validate this point.
Regarding the different examples provided, we like the Belgian and Dutch indicators shown in the discussion paper, but do not have any strong preference as long as the indicator includes at least five risk categories. Consumer testing should be used to determine which presentation enables investors to choose the most appropriate product.
We do not like however the indication typically lower rewards / typically higher rewards" in the UCITS KII synthetic indicator as it appeals to retail investors' greed from a behavioural perspective, which is problematic especially in the current context of very low interest rates."
13: Do you have any views, positive or negative, on the different examples for presentation of performance scenarios? Please outline advantages and disadvantages, and provide any other examples that you are aware of that you think would be useful.
We strongly favour the example with three scenarios within one graph, as it is clear and also shows the impact of an early redemption through displaying the evolution of the pay-out over time.As a second option the example presenting probability in a number of frequencies is also clear and provides a more detailed picture at one point in time, even though it does not display the impact of the holding period. In turn this makes the question of the time frame used for the performance scenarios more acute.
Again we believe that the ultimate choice should be based on consumer testing.
14: Do you have any views on possible combinations of a summary risk indicator with performance scenarios?
We do agree that options 2A, 2B, 2C and 3B are the most fruitful for further exploration during the consumer testing.Consistent with our earlier comments, we favour 2B but believe that consumer testing should determine the final choice.
27: In terms of a possible breakdown of costs, are you aware of cost structures for which a split between entry or exit costs, ongoing costs, and costs only paid in specific situations or under specific conditions, would not work?
We believe that all three risks could be integrated if the first two are based on VaR methodologies and if the impact on VaR of the fluctuations of the bid-offer spread is used for the third one.31: Do you consider that the criteria set out in recital 18 are sufficiently clear, or would you see some merit in ESAs clarifying them further?
We consider that there would be some merit in ESAs clarifying the criteria set out in Recital 18 further.Finance Watch proposed some clarifications and examples in our policy paper Product rules for packaged retail products: why, when, how? " (pages 16-17).
[The PRIIP] invests in underlying assets that are not commonly invested in by retail investors:
We support the proposal to take inspiration from UCITS Article 50.
[The PRIIP] uses a number of different mechanisms to calculate the final return of the investment, creating a greater risk of misunderstanding on the part of the retail investor
We support using ESMA's opinion on types of complex products and tailoring it to fit the specific context of the PRIIPs Regulation.
The investment’s pay-off takes advantage of retail investor’s behavioural biases, such as a teaser rate followed by a much higher floating conditional rate, or an iterative formula
We support further work by the ESAs to develop a common understanding and identify the mechanisms taking advantage of such biases.
As discussed in our policy paper , a number of such mechanisms have already been identified by national supervisors such as the UK's FCA and could be used as a basis (e.g. teaser rate, products designed to frame consumer choice in a potentially misleading way, complex cost structures and claim procedures, “gimmick” packaging features, use of product names implying greater safety than possible etc.).
Additionally, as the Joint Committee recognizes, the Level 1 text lists three criteria that “in particular” should be taken into account to qualify a product as not simple and difficult (for retail investors) to understand. In fact, the original list of criteria as proposed by the Parliament listed 6 features, with the alert being displayed “if one or more” of the criteria would be met. The list was reduced to 4 features in the report as adopted by ECON and the Parliament plenary, and to 3 features during the trialogue phase. However, during that process it was understood that the 3-4 features would be indicative and hence the wording “especially if” was agreed during the final trialogue on 31 March 2014. During legal-linguistic verification, this word was changed to “in particular”, and hence we believe that the list should indeed be interpreted as non-exhaustive.
The above means that national supervisors will have discretion to extend the list of criteria that require the publication of the comprehension alert. As we share the ESAs' concern that this may lead to a lack of harmonised application of the comprehension alert in different member states, we recommend considering further (Level 3) guidance from the ESAs in this area."
34: Do you agree that general principles and as necessary prescribed statements might be needed for completing this section of the KID?
No. The Level 1 text requires the Joint Committees to translate the policy objectives of the Level 1 text into prescriptive (Level 2) Technical Standards, not (Level 3) Guidance.Consumers benefit from uniform simple messages about the objectives of an investment product, and we do not agree that it is difficult for manufacturers to explicitly disclose the objective of a PRIIP.
35: Are you aware of other measures that might be taken to improve the quality of the section from the perspective of the retail investor?
In addition to the description of objectives and the means of achieving them, we would find it extremely useful to have as well one sentence describing in simple terms the implicit underlying market view taken by an investor purchasing a particular product. Such a sentence would ensure that the investor fully understands what view he is taking.Typically such a sentence could be as follows: “an investor purchasing this investment product believes that EU telecom companies stocks will rise over the next five years and NEVER decline by more than 40% at any time during that period.”
The reason for this is that there is a huge difference between understanding how a financial product works and being able to assess the risks attached or understanding the implicit market view taken.
This would therefore contribute greatly in our view to retail investors making informed decisions, as it would highlight clearly the view that they are taking and might make them think again if they do not share this view.