CSA Mutual Fund Risk Classification Methodology for Use in Fund Facts and ETF Facts - Proposed Amendments to NI 81-102 Investment Funds and Related Consequential Amendments

Part I

Background

Part II

Comments on the 2013 Proposal

Part III

Issues for comment

Part IV

Other proposals from commenters

Part V

List of commenters

General comments

Many commenters provided broad support for the CSA's efforts in developing a standardized risk classification methodology, including the objectives and principles set out in the 2013 Proposal.

We thank all commenters for their feedback.

We are proceeding with the Proposed Methodology with proposed rule amendments aimed at implementing the Proposed Methodology for use by conventional mutual funds in the Fund Facts and exchange-traded mutual funds (ETFs, together with conventional mutual funds, mutual funds) in the proposed ETF Facts.{1}

One commenter, The Investment Funds Institute of Canada (IFIC), acknowledged that although the risk classification methodology developed by IFIC (the IFIC Methodology) was developed only for IFIC's members, they supported making it publicly available for use by non-members as well.

From our research, we know that the IFIC Methodology is the predominant risk classification methodology currently used by fund managers. Our Proposed Methodology was informed by the feedback we received on the 2013 Proposal. We note that the Proposed Methodology is consistent with the IFIC Methodology in many respects, including the use of standard deviation (SD) as a risk measure, a five-band risk scale, and the SD ranges for the risk bands. We believe this should minimize the changes in investment risk levels for funds resulting from the implementation of the Proposed Methodology.

1. As a threshold question, should the CSA proceed with (i) mandating the 2013 Proposal or (ii) adopting the 2013 Proposal only as guidance for IFMs to identify the mutual fund's risk level on the prescribed scale in the Fund Facts?

Several commenters emphasized that any risk classification methodology developed by the CSA should be mandated so that investors can readily compare funds knowing that the investment risk levels of mutual funds are determined using a standardized risk classification methodology. One commenter noted that this would assist investors in making informed investment decisions.

The CSA have decided to move forward with a mandated standardized methodology. In addition to written comments received, the majority of experts we consulted with in Fall 2013 also recommended the use of a standardized risk classification methodology in order to level the playing field between mutual funds, and to eliminate arbitrage. Adopting a standardized risk classification methodology would achieve the objective of comparability across asset classes and mutual fund products.

Are there other means of achieving the same objective than by mandating the 2013 Proposal, or by adopting it only as guidance?

We request feedback from IFMs and dealers on what a reasonable transition period would be for this.

One commenter believed that requiring the adoption of a more objective and uniformly applied metric such as SD will help reduce and eliminate "arbitrage" whereby some fund managers may determine the investment risk level by using subjective factors and giving a product a lower rating than it may otherwise warrant based on a more objective assessment.

While supporting a risk classification methodology prescribed by the CSA, one commenter suggested that where the chosen standard is impractical to implement or when it would lead to meaningless or misleading results, exemption requests should be considered by the CSA.

Several commenters also commented that it is beneficial for Canadians to have all mutual funds evaluated on a consistent standard. However, these commenters recommended that the CSA consider adopting the current IFIC Methodology as the new mandatory standard. This would accomplish the CSA goal of ensuring consistent determination of investment risk levels across all mutual funds and also have a limited impact on existing Canadian investors and the industry. This would enable a shorter transition period.

As mentioned above, the 2013 Proposal has several features that are consistent with the IFIC Methodology, including the break points for the various risk bands. We expect that this will help reduce any transition period following the implementation of the Proposed Methodology. We note that the IFIC Methodology, as currently constructed, allows for significant use of discretion by fund managers and has not been consistently applied by fund managers in rating their mutual funds.

Two commenters suggested that the IFIC Methodology is widely used by the vast majority of the industry and is easily understood by investors, and therefore, the IFIC Methodology should be adopted to minimize any impact on investors.

Along the same lines, one commenter suggested that the CSA rule should mandate use of a single methodology which is managed by an industry group with appropriate knowledge and experience to meet the objectives (expanded to include investor interest) as set out in the CSA proposal. The commenter believed that management of guidance relating to the IFIC Methodology through IFIC's Fund Risk Classification Task Force could be expanded to include representatives from different industry segments, with the CSA as observers when the methodology itself is discussed annually.

One commenter urged the CSA to consider the Committee of European Securities Regulators (CESR), now European Securities and Markets Authority (ESMA), risk classification methodology for adoption in Canada.

In developing the 2013 Proposal, the CSA analyzed and considered both the IFIC and CESR methodologies. The 2013 Proposal has been amended based on the feedback received and, we believe, best fits the criteria and objectives as outlined in it. It should be noted that the European summary document and risk scale have significant differences compared to our summary documents. In our view, the Proposed Methodology best reflects the reality of our mutual fund market which allows for comparability across mutual funds.

Several commenters believed that the CSA should adopt high level principle-based guidance with respect to risk classification rather than mandate the 2013 Proposal.

The CSA believes that a standardized risk classification methodology is needed to enable investors to make meaningful comparison between mutual funds. We believe that a standardized risk classification methodology will benefit all mutual funds with greater transparency and consistency. It is our view that high-level principle-based guidance could not achieve either of these objectives, as it would allow room for potential manipulation.

In one commenter's view, if the risk rating is not subject to fund manager discretion then it should only be guidance.

One commenter did not recommend adopting the 2013 Proposal as guidance for fund managers, as it would co-exist with the currently used IFIC methodology, leading to non-comparability of information in the Fund Facts.

2. We seek feedback on whether the 2013 Proposal could be used in similar documents to Fund Facts for other types of publicly-offered investment funds, particularly ETFs.

Several commenters were of the view that the same risk classification methodology should apply to all investment funds to ensure a level-playing field for all products.

We are proposing that the Proposed Methodology be used both for exchange-traded mutual funds and conventional mutual funds.

For ETFs, what, if any, adjustments would we need to make to the 2013 Proposal?

Some commenters asked how alternative funds, closed end funds, leveraged ETFs or structured products' risk rating would be determined. These commenters questioned that if these non-mutual fund products come out as high risk from a volatility perspective, would comparisons by retail investors be meaningful or misleading? These commenters question whether volatility alone is a sufficient measure of risk for these types of products. There may be high-risk mutual funds that are significantly less risky than a high-risk closed-end fund or alternative fund but this may not be apparent, if they are all bunched in the same risk category. Some commenters suggested that the limitations of volatility risk will likely become evident when trying to expand summary disclosure to other types of funds.

We note that alternative funds, closed end funds and structured products are not currently required to produce a Fund Facts or an ETF Facts, and therefore, are not required to determine their investment risk level. Therefore, the Proposed Methodology will not apply to such products. Should the disclosure requirements for these non-mutual fund products change, the CSA would consider the applicability of the Proposed Methodology to such products.

For instance should standard deviation be calculated with returns based on market price or net asset value per unit?

Several commenters favoured using market price data rather than net asset value (NAV) in calculating SD for ETFs since it is more reflective of the returns investors are likely to realize.

The CSA conducted research on this issue to assess whether there are significant differences in the investment risk level of a mutual fund if market values are used versus NAV. While a very small minority of ETFs provided a different risk rating by using market value versus NAV, we note that the larger issue the CSA encountered was consistent availability of market values for thinly traded ETFs or for the advisor series of ETFs. Given the lack of consistent market value data for ETFs, the CSA are proposing that NAV be used to determine investment risk level.

Two commenters submitted that whether SD is best measured based on market price or NAV would be best determined by a focussed investigation. One of these commenters urged the CSA to include ETFs in the study before publishing any proposals.

Using NAV to determine investment risk level also allows for consistency with performance reporting and continuous disclosure requirements for mutual funds.

3. We seek feedback on whether you agree or disagree with our perspective of the benefits of having a standard methodology, as well as whether you agree or disagree with our perspective on the cost of implementing the 2013 Proposal.

The vast majority of commenters who answered this question agreed with the CSA's perspective on the benefits of having a standard risk classification methodology as it will provide consistency and transparency of disclosure and improved comparability of different mutual funds.

We agree that a standardized risk classification methodology will enhance transparency and ensure comparability between mutual funds. We have made a number of changes to the 2013 Proposal specifically in response to the comments received regarding the impact on dealers. We have retained the five-category risk scale currently used in the Fund Facts, used SD as the risk indicator and our proposed risk band break points are consistent with those used by the IFIC Methodology. We believe these changes to the 2013 Proposal will minimize the cost of implementation for both fund managers and dealers.

Some commenters estimated that many fund managers will have a significantly high percentage of their mutual funds moving to a higher risk classification under the 2013 Proposal, resulting in significant impact for dealers and investors.

Two commenters added that the cost to fund managers and dealers would be minimized if the IFIC Methodology is adopted since most firms already calculate and review the risk associated with their product in accordance with this methodology.

A few commenters who agreed with the benefits of having a standardized risk classification methodology suggested that the cost incurred by fund managers is not expected to be significant if current risk categories and risk band breakpoints are not changed. This is because dealers would not have to amend their processes and systems technology to accommodate changes. Changes in the risk classification of funds, however, would require dealers to conduct client account reviews, re-paper client accounts and/or change client portfolio allocations.

4. We do not currently propose to allow fund IFMs discretion to override the quantitative calculation for risk classification purposes. Do you agree with this approach?

Several commenters agreed that fund managers should not be allowed to override the quantitative calculation for risk classification purposes. Two of these commenters suggested that if only a quantitative metric is used to determine the investment risk level, the CSA should allow fund managers discretion to move their risk classification higher only.

After considering the comments received, the CSA recognize that circumstances could give rise to the need for consideration of qualitative factors in addition to the quantitative calculation in determining the investment risk levels of mutual funds. Therefore, the Proposed Methodology contemplates the use of discretion to classify a mutual fund at a higher investment risk level.

Should we allow discretion for IFMs to move their risk classification higher only?

A few commenters explained that not allowing the use of qualitative factors for the purposes of determining investment risk levels was advantageous as discretion can lead to misleading ratings and defeat the goal of comparability and transparency. One commenter added that if truly extraordinary circumstances prevail, some explanatory disclosure should be allowed.

However, the CSA are of the view that there should be no discretion to classify a mutual fund into a lower investment risk level. We consider that a mutual fund should be classified, at a minimum, at the investment risk level determined by its SD.

Several commenters were of the view that other types of risk, both measurable and non-measurable, may exist. The commenters believed fund managers must retain their discretionary power to classify an investment fund either higher or lower than the risk classification indicated by quantitative results. Doing so allows a fund manager to make full, true and plain disclosure of all material facts relating to the investment funds being offered. By removing discretion completely, the 2013 Proposal removes the responsibility of fund managers to consider other factors that could affect the risk of a fund, and thus reduces the responsibility to disclose all risks. One of the commenters added that the prospectus and Fund Facts impose civil liability so it is crucial that a fund manager is comfortable with the investment risk level assigned to a particular fund. Some commenters believed that a fund manager can document the reasons for deviating from the numerical SD calculation where they do so.

One commenter supported the inclusion of a qualitative element which could be monitored by a third party, in conjunction with industry input and participation.

Another commenter told us that it was important that fund managers be provided with discretion when determining the investment risk classification of funds in order to maintain consistency year over year. The commenter added that fund managers should be prepared to defend their use of discretion if it is questioned by the CSA.

5. Keeping the criteria outlined in the introduction above in mind, would you recommend other risk indicators?

Approximately two thirds of the commenters agreed with the use of SD as a comparable measure of risk for the purposes of a risk classification methodology. SD's simplicity, objectivity and relevance in measuring volatility risk are shared by the commenters. Its applicability to a large range of funds was also commended.

The CSA propose to keep SD, which measures volatility of past returns of the mutual fund, as the risk indicator for the Proposal Methodology. We are of the view that given the available alternatives and the known data obstacles, SD is still the best general risk indicator and one that is useful as a first test to measure overall risk. Our analysis of data from the Canadian fund marketplace revealed that there were relatively few cases where alternative risk indicators signaled a higher risk rating than that indicated by SD. We also note that most risk indicators will tend to underestimate risk where the probability of event risk (i.e. unforeseen event) is high.

If yes, please explain and supplement your recommendations with data/analysis wherever possible.

While commenters generally supported the use of SD, some remained concerned with over-simplifying mutual fund risk to a single, quantitative measure. The commenters suggested that when asked about risk, many investors indicate their greatest concern is the risk of loss of capital, which is not captured by SD.

Before the CSA decided on SD as its preferred risk indicator, we conducted a thorough study of 15 other indicators. The other indicators studied included, among others, risk/return indicators, (such as the Sharpe Ratio, the Information Ratio and the Sortino ratio), tail risk indicators (such as Value at Risk (VAR), CVAR) and performance indicators (such as worst period). Our study included an assessment of how well each of these indicators met our principles for the development of the Proposed Methodology. Further, we also assessed if any of these indicators added further value as a secondary indicator in addition to using SD as a primary indicator.

To perform this analysis, we looked at data from mutual funds that were available in Canada from 1985 to 2013. We noted that these indicators tended to have significant correlation with SD. In other words, if VAR, as an example, indicated high risk for a particular fund, SD would have a similar higher risk indication. In only a small minority of instances (less than 2%) did SD tend to underestimate risk relative to another indicator such as VAR. Even in such instances, these funds tended to be small/mid cap equity and resource/precious metals equity funds, which already tend to be classified in the Medium to High or High risk category based on the SD calculation. We, therefore, concluded that SD did as good a job as any other indicator, and the additional complexity and regulatory burden associated with adding a secondary indicator was not justified.

A few commenters opposed the use of SD as an indicator of risk disclosure in the Fund Facts. They felt that SD is not easily understood in practical terms by most retail investors. They wondered if retail investors will understand that a fund with a high SD does not necessarily mean that such a fund is worse than another with a low SD.

Since the creation of the Fund Facts, SD has been widely used to determine the investment risk level of a mutual fund on the risk scale in the Fund Facts. While investors may not be able to understand the mathematical calculation of SD, there is a plain language description of volatility in the Fund Facts. The investment risk level, along with other key information in the Fund Facts, such as the suitability section will help investors make an informed investment decision.

Several commenters believed that SD requires some knowledge of mathematical statistics to be employed effectively for informed decision making. Such approach is much too complex to be used by retail investors, no matter how well described in plain language.

Further, in the Fund Facts, under the risk scale, there is a cross reference to the Risk section of the mutual fund's simplified prospectus for more information on risks.

Another commenter was concerned that SD is an insufficient, inappropriate and not well-understood measure of risk. Additional descriptions of risk exist and are preferable as they propose a table/graph of worst-case and best-case historical return scenarios that can be used to demonstrate fund volatility. According to this commenter, the Fund Facts' disclosure of volatility is presented and used as though it gives an indication or assurance of future variability/risk. The commenter encouraged the CSA to do exhaustive cognitive and behavioural testing to determine what patterns of variation a risk-averse investor would view as risky before finalizing the statistical models, the classifications and the ranges that have been proposed. In the commenter's opinion, investors understand risk in terms of potential dollar losses in their portfolio more easily than percentage returns. In the commenter's experience most investors can understand graphs and tables far more readily than calculations such as SD.

The CSA disagrees with the commenter. Past volatility is not presented in the Fund Facts as being an assurance of future variability. Under the section "How risky is it?" in the Fund Facts, it states "This rating is based on how much the fund's returns have changed from year to year. <<It doesn't tell you how volatile the fund will be in the future. The rating can change over time. A fund with a low risk rating can still lose money>>."

Under the same section, there is a cross reference to the Risk section of the mutual fund's simplified prospectus for more information about the risk rating and specific risks that can affect the mutual fund's returns.

According to one commenter, SD on its own does not tell us anything about the uncertainty of price movements (be it their size or their probability of occurring) or the uncertainty of events surrounding price movements, or whether it is a good or a bad risk to assume. Therefore relying on SD as the sole information point about risk does not inform the investor about the actual range and impact of outcomes that could affect them.

Two commenters were of the view that looking at volatility risk alone can be misleading and lead to sub-optimal decisions for the investor. As a result, some risk/return metric disclosure should be added as a supplement to any type of risk disclosure. Metrics such as Sharpe ratio and Information ratio would provide additional clarity to how effectively fund managers use risk and how consistent their returns are. These commenters added that the Sharpe ratio and the Sortino ratio are far more meaningful as they measure risk adjusted returns. The Sharpe ratio allows an investor the ability to quantify an investment's risk relative to its investment performance in order to decide if a financial product is worth the risk. One of these commenters noted that the Sortino ratio is a more meaningful measure of investment risk than SD as the Sortino ratio is similar to the Sharpe ratio, but its denominator focuses solely on downside volatility, not overall volatility. It is only downside volatility that is relevant and unwanted. This is a serious flaw in the calculation of both SD and the Sharpe ratio as a measure of risk. The Sortino ratio is a more meaningful measure of investment risk than SD.

Please see response above which describes the CSA's analysis in regard to consideration of other metrics.

The commenter recommended that investment risk levels be measured based on portfolio holdings, thus reflecting the inherent risks. Should the CSA proceed with mandating a standardized risk classification methodology, the commenter strongly recommended that it be based on a blend of measures that includes Conditional Value at Risk (CVAR) and a holdings-based approach. The commenter believed that the use of the SD measure as the sole measure of risk does not serve the best interests of the investors.

Please see response above which describes the CSA's analysis in regard to consideration of other metrics.

6. We believe that standard deviation can be applied to a range of fund types (asset class exposures, fund structures, manager strategies, etc.).

Several commenters agreed that a uniform measure should be applied across all investment funds.

We thank commenters for their feedback.

Keeping the criteria outlined in the introduction above in mind, would you recommend a different Volatility Risk measure for any specific fund products?

Two commenters submitted that given the structured nature of target date funds, balanced funds and T-class series of securities, a different approach to articulating risk is required for these types of funds.

Please supplement your recommendations with data/analysis wherever possible.

In regard to target date funds, commenters indicated that one of the associated risks is a premature movement to a safe mode (a "triggering event") which happened in 2008 -- such a risk is not captured by SD. Further, life cycle funds are designed such that their risk level changes over time, so a backward looking risk measure may not be a suitable indicator of product risk as it may overstate the risk of the fund at a point in time.

In order to address concerns relating to overstatement of investment risk levels for target date funds, we performed an analysis of the volatility profile of current target date funds. The analysis demonstrated that target date funds closer to their target date did indeed have lower SD, however, the difference in SD over the life cycle of target date funds was relatively small owing primarily to the inherent diversification attributes of products. Thus, we expect that many target date funds will remain in the same risk band over the course of their existence and those that do shift will not shift by more than one risk band, and even then very slowly. Therefore, the CSA did not propose a change to the Proposed Methodology since overstatement of risk for target date funds was not supported by the data studied.

In regard to balanced funds, commenters noted that constant changing of asset mix can be a challenge in regard to risk classification. Similarly, some commenters pointed to tactical asset allocation funds as a challenge for the proposed risk classification methodology since the underlying statistical distribution is constantly changing for such funds.

For balanced funds and T-series of securities, the 2013 Proposal allows for discretion to use a reference index as a proxy for missing information that best fits the risk profile of such funds. The reference index can be a single index or a blend of indices that best fits the risk profile, and therefore, should allow an index to be customized to the risk profile of the fund.

Similarly, commenters also pointed to T-series of securities that return capital each month, suggesting that finding an appropriate index for the purposes of backfilling information may be difficult. Further, such mutual funds run the risk of disintegration if payouts are too steep, and such a risk is not captured by SD.

The Proposed Methodology requires that the investment risk level of a mutual fund be determined by using the oldest series of the mutual fund, unless the oldest series has a different profile or materially different terms associated with it. As such, where appropriate, the investment risk level of currency hedged series of a mutual fund should be determined separately if it is materially different to the other series of the mutual fund.

Commenters also suggested that currency hedged funds complicate return distribution profile and fund behavior/volatility, thus a different approach may be needed for currency hedged funds, such as a separate SD calculation for the hedged and unhedged series of a mutual fund.

One commenter noted that ETFs and exempt funds by their nature are different products. The commenter supported investigating the possibility of using a different volatility risk measure for specific fund products.

As noted above, we are proposing that the Proposed Methodology be used both for exchange-traded mutual funds and conventional mutual funds.

One commenter agreed that a risk classification methodology that is based on SD of fund returns is a good measure of a fund's risk. However, fund managers should have the flexibility to supplement SD with other measures that may be more tailored to the specific fund. A good measure for a fixed income fund, for example, would be duration, which is a measure of sensitivity to interest rate risk, added this commenter. Another possible measure, for a fund that uses derivatives particularly, would be VAR.

Please refer to our responses under question 35 in regard to applicability of other risk measures in addition to SD.

7. We understand that it is industry practice (for IFMs and third party data providers) to use monthly returns to calculate standard deviation. Keeping the criteria outlined in the introduction above in mind, would you suggest that an alternative frequency be used?

Commenters agreed that using a mutual fund's monthly returns is appropriate. Commenters added that monthly data is traditionally used to assess risk and return data in the mutual fund industry.

Given the feedback from commenters, the CSA are keeping the monthly returns with reinvestment of all income and capital gains distributions for the Proposed Methodology.

Please specifically state how a different frequency would improve fund risk disclosure and be of benefit to investors.

Please supplement your recommendations with data/analysis wherever possible.

8. Keeping the criteria outlined in the introduction above in mind, should we consider a different time period than the proposed 10 year period as the basis for risk rating disclosure?

Several commenters agreed with the proposed 10 year period as the basis for risk rating disclosure. One commenter added that a 10 year period has the effect of attenuating sudden changes in financial markets and helps smooth out extreme fluctuations which are often temporary.

After reviewing fund data for the Canadian fund marketplace, we are of the view that the use of ten-year performance returns is preferable to both shorter (3, 5, 7 years) and longer time periods (15, 20, 25 years) as it strikes a reasonable balance between indicator stability and data availability.

Please explain your reasoning and supplement your recommendations with data/analysis wherever possible.

Although one commenter supported the use of longer-term performance data to calculate SD, the commenter suggested that this be modified to 10 years or as far back as required to include at least one bear market for the mutual fund or its relevant benchmark.

We also note that the CSA studied data of available mutual funds and various indices using varying time periods ranging from three, five, seven, ten and fifteen years for the calculation of the SD. We noted that three, five and seven year SD results caused frequent risk band changes for a number of funds resulting in significant costs for fund manufacturers as well as dealers. Compared to such time periods, a 10 year SD calculation was a more stable indicator of risk. We note that moving from a 10 year SD calculation to a 15 year SD calculation only provided minimally increased stability as a risk indicator, and any benefits from moving to a time period longer than 10 years would be offset by the costs of gathering data for a longer time period. We also note that a 10 year time period typically tends to catch at least one downturn in economic and financial markets.

One commenter agreed with the proposed 10 year period as the basis for comparison of SD across mutual funds. However, the commenter was of the view that a 10-year period would be insufficient for measuring risk of loss. There are long periods of time where capital markets have delivered strong performance with limited downside. While a rolling 10-year measurement period will not significantly impact the SD calculation, it could significantly impact the worst and best returns. For risk of loss to be a stable indicator, it requires a static start date, with as long a time period as possible (for example, starting from 1960).

Some commenters disagreed with the use of a 10 year time period for the purposes of the SD calculation. One commenter noted that the average lifespan of a mutual fund is less than 6 years, while studies indicate that the average holding period of a mutual fund is less than 5 years and shrinking. This indicates that a typical investor will not experience the smooth, consistent ride that a 10 year SD implies, but will experience the swings in volatility that occurs over a 5 year period. The commenter conceded that using the 10 year period will ensure that mutual funds are not frequently switching risk categories.

In regard to comments about the average life of a mutual fund and the average holding period of a mutual fund, we note that the investment risk level is intended to capture the volatility risk of a particular mutual fund and a particular asset class rather than providing an assessment of the risk profile of an average mutual fund investor.

One commenter felt that the use of a 3-year annualized SD model would decrease the ability of funds to obfuscate their risk rating and allow for better comparability across all mutual funds, as more funds would possess this complete return history. Another commenter suggested that the CSA should consider whether it is better to use a 7-year SD if this presents fewer incidences of needing to use a reference index as a proxy and will, therefore, be subject to less manipulation.

One commenter thought that using a 10-year history to calculate the SD for an investment fund may result in an investment fund being classified as more volatile than it actually is if there are two volatile periods i.e. at the beginning and at the end of the 10 years. The commenter believed that using three-to-five-year historical data would be the appropriate timeframe as this represents the average time that an investor holds securities of an investment fund.

Several commenters did not believe that a 10-year annualized SD provides any more information than the 3 or 5 year annualized SD presently prescribed under the IFIC methodology. These commenters recommended adopting 3 or 5 year annualized SD similar to the IFIC Methodology.

To the best of another commenter's knowledge there is no research indicating that 10 years is a better indicator of a market cycle versus 5 years or 15 years, other than that the longer periods smooth results.

One commenter noted that requiring the presentation of a 10 year measure of volatility (real or simulated) is contrary to the CFA Institute's Global Investment Performance Standards (GIPS). The commenter suggested that rather than selecting one risk category for a fund, the volatility of the fund be presented over time in graph format by showing, for each period, the annualized three year SD. This commenter recommended shortening the period to 5 years, similar to the CESR Guideline.

We note that the purpose of the GIPS presentation is entirely different from the purposes of presentation of risk classification level in the Fund Facts or ETF Facts. GIPS performance presentation aims to ensure fair presentation of investment performance results of money managers, rather than an assessment of the risk level of their portfolios.

9. Keeping the criteria outlined in the introduction above in mind, should we consider an alternative approach to the calculation by series/class?

A few commenters agreed that a consistent approach should be applied across all series/class of a mutual fund.

Our analysis concluded that the variance of the SD calculation is small across series/classes of securities of the same mutual fund. For this reason, and after considering the comments received, we are not requiring that the investment risk level be determined for each series/class of securities of a mutual fund, unless a series/class of securities possesses an attribute that could result in a different investment risk level than that of the mutual fund. In such instances, the investment risk level should be determined for that particular series/class of securities. An example of such an instance would be a currency hedged series/class of securities of a mutual fund which could have materially different performance returns relative to the other series of the mutual fund which may result in a different investment risk level.

Please supplement your recommendations with data/analysis wherever possible.

One commenter did not believe that it is necessary to apply the 2013 Proposal to individual series/classes of a mutual fund. Each series/class of a mutual fund has identical fund holdings and therefore bears equivalent levels of risk. While it is true that returns vary by series/class, differences in SD are slight to non-existent.

Several commenters submitted that the fund manager should use the total returns of the "oldest" mutual fund series/classes as the basis for his/her volatility risk calculation across all the mutual fund series/classes having the same strategy as the volatility risk remains the same. Two of these commenters added that this should be the case unless an attribute of a particular fund series/class would result in a materially different level of volatility risk (e.g. currency hedging), in which case, the total returns of that particular mutual fund series/class must be used.

One commenter told us that risk should be calculated and reported separately for different series of a mutual fund's units (for example, D and F class series) given that the greater the fees, the greater the risk of loss while SD does not change.

10. Keeping the criteria outlined in the introduction above in mind, do you agree with the criteria we have proposed for the use of a reference index for funds that do not have sufficient historical performance data?

A few commenters agreed with the use of a reference index in the absence of sufficient historical statistical information. One commenter not only agreed with the use of a reference index for the purpose of backfilling missing data but suggested that funds that have a 10 year history should provide data corresponding to a reference index similar to their funds. In so doing, investors could compare a fund's volatility with the volatility of its reference index.

The CSA are aware that the majority of mutual funds do not have 10 years history required for the Proposed Methodology. To address this issue, we have proposed the use of a reference index as a proxy for the missing data. The Proposed Methodology sets out criteria for what constitutes an appropriate reference index to be used as a proxy for the purposes of backfilling missing data history.

Are there any other factors we should take into account when selecting a reference index?

Please supplement your recommendations with data/analysis wherever possible.

One commenter was of the view that using a reference index is not an appropriate method of representing true expected volatility of any mutual fund and may lead to unintended consequences. When the performance of a reference index is compiled with the historical returns of a mutual fund, it does not allow investors to determine if the fund manager's active management style adds to the volatility of the fund or whether that is a function of its reference index. The commenter believed that permitting a fund manager to choose a reference index as a proxy will insert a measure of uncertainty and discretion into the calculation. In order to reduce some of the discretion, the commenter recommended that if use of a reference index as a proxy is permitted, fund managers should also be required to perform the calculation based only on the actual returns of the mutual funds and show that information alongside the reference index, and explain (if there is a difference) how the mutual fund would fit in a different risk band if the actual performance history and not using the reference index as a proxy for the missing returns over a 10 year period.

The Proposed Methodology requires the selection of a reference index that reasonably approximates the volatility and risk profile of the mutual fund. The Proposed Methodology also sets out criteria for selecting and regularly monitoring the appropriateness of the reference index. We do not propose to add the suggested data points to the Fund Facts at this point as this is only likely to add confusion, in particular, for retail investors.

Two commenters suggested that the use of a reference index is contrary to every other CSA publications, particularly CSA Staff Notice 31-325 Marketing Practices of Portfolio Managers issued July 2011 (a successor to OSC Staff Notice 33-729 Marketing Practices of Investment Counsel/Portfolio Managers issued November 2007). In both notices, the use of hypothetical or simulated performance data, especially for retail investors, is basically prohibited. Only actual returns are to be presented. It is also noted that under no circumstances are hypothetical and actual returns to be linked, which the 2013 Proposal specifically requires. The prohibition on hypothetical data is due to the various risks and inherent limitations in using such data, as outlined in the Notices. Consequently, the use of a reference index as a proxy for returns over a 10 year period as if they were achieved by the mutual fund and linking them to actual returns, is contrary to established CSA policy. The generation of a hypothetical or simulated risk profile, utilizing a linkage of theoretical and actual returns, is also prohibited by the CFA Institute GIPS.

The CSA believe that the use of a reference index data in determining the investment risk level of a mutual fund is not contrary to previous CSA publications on the use of hypothetical or simulated performance data. The use of reference index data in the Proposed Methodology is limited to determining the investment risk level of a mutual fund which is disclosed in the Fund Facts or ETF Facts. The reference index is not used as a representation of a mutual fund's performance but rather it acts as a proxy for missing data in determining its investment risk classification using the Proposed Methodology.

Two commenters asked that the CSA provide greater clarity around what can be used as a reference index, for instance whether fund managers may use blended indices and if so, whether such use must be disclosed in the mutual fund's prospectus. It should also be clarified in what circumstances, if any, a change in reference index from what was originally disclosed would constitute a material change.

The Proposed Methodology allows for the use of blended indices and requires that if the reference index has changed since the last prospectus, the prospectus provides details of when and why the change was made.

Several commenters suggested that the reference index be consistent with the broad-based market index chosen for the Management Report of Fund Performance (MRFP). Applying different criteria for the MRFPs and the fund's risk classification will create confusion for both investors and dealers added another commenter.

The same index or indices used in the MRFP of a mutual fund can be used to determine its investment risk level if the index or indices reflect the risk profile of the fund and meets the criteria for an appropriate reference index as outlined in the Proposed Methodology.

Two commenters agreed that fund managers should have the discretion to select an appropriate reference index to increase the information set of a fund to 10 years. These commenters would, therefore, extend this consideration to also allow using imputed data in situations where a fund's past returns are not representative of the fund's current attributes due to material and intentional changes to the fund. For example, if a mutual fund's securityholders vote to modify the fundamental investment objectives of a mutual fund, such that the returns of the fund would behave differently than it has previously, essentially making it a new mutual fund. One of these commenters also wanted to caution the CSA that determining an appropriate reference index may be difficult for mutual funds with volatility of returns that are different than any existing reference index.

We agree with the comments made and have made some changes to the 2013 Proposal to address instances where there has been a fundamental change in the investment objectives or a reorganization of a mutual fund.

One commenter noted that there is no perfect solution to choosing a reference index and that the investment objectives of some mutual funds are so flexible and unique that none of the widely available benchmarks capture the mutual fund's exposure or strategy. Two commenters were of the view that a mutual fund's returns may not be highly correlated to the index because of the mutual fund's active investment strategies The 2013 Proposal requires a reference index to meet each of the stated criteria which prove particularly difficult for innovative mutual funds where risk management is held out as a defining feature of the mandate, such as low volatility and target return funds.

According to the criteria for a reference index set out in the Proposed Methodology, the returns of the reference index should be correlated to the returns of the mutual fund, rather than replicate the returns exactly. As such, we believe there are sufficient reference indices available that can serve as a proxy for the risk profile of actively managed funds.

Another commenter proposed that the CSA should consider Canadian Investment Funds Standards Committee (CIFSC) category-based benchmarks as potential proxies because they are better proxies for the investor experience than market-based benchmarks.

Fund managers have discretion in their selection of the reference index as long as the reference index appropriately reflects the risk profile of the fund's investment objectives and meets, among other things, the criteria outlined by the CSA in regard to what is an appropriate reference index.

One commenter requested clarification on the conditions that the indices be "widely recognized" and "publicly available". On the criterion of "publicly available", the commenter noted that very few index publishers issue monthly data or make the SD of index returns available to the public free of charge. The commenter also noted that many fund types, such as sector funds, real estate funds, high yield funds and floating rate debt funds, would generally find the most suitable proxies among indices that are neither widely recognized nor whose data is publicly available.

In response to comments, we have removed the requirement that the reference index be widely recognized and publicly available in all instances.

Two commenters believed there may be some concerns surrounding the practice of the fund managers selecting their own reference indexes as fund managers may aim keep the risk rating of their fund at a certain level. In such instances, the fund manager could choose an index with the lowest possible investment risk level while abiding by the lax criteria put forth by the CSA. Having a third party, such as data providers or industry participants, select the reference index on behalf of the fund manager would eliminate the conflict of interest. One of these commenters also had concerns as to whether or not the CSA has the means to effectively monitor index selection to ensure the chosen benchmarks accurately reflect the potential volatility of a mutual fund.

We believe that the requirement to disclose the chosen reference index in a mutual fund's prospectus allows for transparency. Where CSA staff have questions around the appropriateness of a reference index, the mutual fund may be the subject of a continuous disclosure review in this area.

One commenter was of the view that certain fund of funds may not have the requisite 10 year history however, the underlying fund may have been in existence for a longer time period. In this case, using the returns of a reference index would not be a meaningful representation of a fund's risk level, rather preference should be given to the performance history of the underlying fund which may have been in existence for a longer time period.

In instances where the underlying fund has a 10 year history, and the top fund's stated investment objectives and strategy is to "clone" that underlying fund, staff may consider allowing, through exemptive relief, the use of the underlying fund's volatility of returns for the purposes of determining the top fund's investment risk level.

Two commenters believed that the consultation paper should have provided details of exactly how costless index returns are to be adjusted in order to link to actual after-fee fund returns to obtain 120 data points where actual data is less than 10 years.

We do not propose that index data be adjusted for fees. We do not believe fees impact volatility of returns to a significant extent.

11. Keeping the criteria outlined in the introduction above in mind,

Several commenters told us that the 2013 Proposal's risk bands and associated risk categories will lead to a large number of mutual funds being re-classified into a higher investment risk level, without any associated change in the mutual fund's risk. According to two of these commenters, between 70% to 80% of their mutual funds would move upwards to a higher investment risk level under the 2013 Proposal. One of the commenters did not believe that it is necessary to have a "Very High" investment risk level as there are very few mutual funds which would be included in this band. A few commenters recommended that the CSA use the same number of risk bands and the same nomenclature as described in the IFIC Methodology to avoid investor confusion and industry disruption.

In response to the concerns expressed by commenters about the change in the risk scale from 5 categories to 6 categories and the associated costs, the CSA are proposing to retain the current CSA five-band risk scale used in the Fund Facts to avoid unnecessary reclassification of mutual funds and suitability reassessments which may be triggered as a result. While our intention in proposing a six band risk scale was to improve the segregation of asset classes across risk bands, we acknowledge stakeholders concerns raised in this regard.

i. Do you agree with the proposed number of risk bands, the risk band break-points, and nomenclature used for risk band categories?

ii. Do the proposed break points allow for sufficient distinction between funds with varying asset class exposures/risk factors?

If not, please propose an alternative, and indicate why your proposal would be more meaningful to investors.

Please supplement your recommendations with data/analysis wherever possible.

One commenter preferred the use of 5 risk categories rather than 6 for the reason that current know your client (KYC) are based on 5 band risk tolerance levels. According to the commenter, losing the symmetry between the KYC classification and the know your product (KYP) investment risk level from the Fund Facts will seem illogical and create confusion for investors and their advisors.

Two commenters noted that under the 2013 Proposal, the majority of mutual funds would be labeled as "Medium-to-High", while typically exhibiting only a fraction of the volatility of the highest risk investments. Given the range of investment options and associated investment risk levels, it is not intuitive that broad-based equity mutual funds, which typically exhibit risk levels consistent with broad markets, would be have a "Medium-to-High" investment risk level.

Several commenters queried whether the additional investment risk level of "Very High" is necessary in light of it extremely limited applicability. One of these commenters urged the CSA to consider an alternate labeling system with investment risk levels ranging from "Very Low" to "High" which would limit unnecessary material change filings, prospectus amendments and suitability reviews which would ultimately be more cost-effective and minimize confusion for investors in this area.

Along the same line, a few commenters questioned why this new risk scale is any better than the current scale, given that it was the CSA that developed the current risk scale (mandated in conjunction with the Fund Facts regime introduced in January 2011). One commenter questioned the meaning of the CSA's explanation that the new investment risk levels will achieve "more meaningful volatility clustering in the fund universe" and also asked how the new risk bands -- including the new sixth band -- achieve this.

One commenter believed that the thresholds have been set somewhat too low; i.e., the proposed bands place mutual funds that the commenter believed should be in a lower risk category into a higher one.

One commenter fundamentally disagreed with the CSA's proposal to fix the risk band break points. The fundamental problem is that values of the ranges were presumably selected to represent the riskiness of specific asset classes over some historical period, but there is no guarantee that the values will continue to do so in the future, as the risk levels of asset classes' change over time. For this reason the commenter favoured a system with floating risk bands. According to two commenters, applying the 2013 Proposal while maintaining the bands and labels from the IFIC Methodology would result in fewer funds requiring re-classification during implementation of the 2013 Proposal. This approach would also significantly reduce the transition time.

One commenter believed that there should be a distinction between mutual funds that claim to offer full principal stability, such as money market funds, and those that offer high but not complete principal stability. The commenter added that there would be a benefit to adopting the same 7 band scheme as the CESR methodology.

For the benefit of the investor and to provide a clearer picture of the actual risk level of the mutual fund, one commenter proposed that rather than increasing the number of risk categories available, the CSA simply require mutual funds to indicate its SD on the risk scale in the Fund Facts. In this manner, an investor would have a more accurate indication of the relative risk level for the mutual fund and an easy way to compare mutual funds with similar mandates, and the need to reclassify investment risk levels and/or increase the number of risk bands is reduced.

Two commenters acknowledged that adopting the 2013 Proposal may result in changes to the investment risk level for some mutual funds. However, the commenters submitted that the need for some reclassification of funds into a different (and more accurate) investment risk level is not a valid reason not to adopt a standardized risk classification methodology.

12. Do you agree with the proposed process for monitoring risk ratings?

The majority of commenters believed that monthly monitoring is excessive and burdensome. Several commenters recommended semi-annual or annual monitoring. Several other commenters recommended that the CSA simply adopt an annual monitoring process that is tied to a fund's annual renewal and that it be aligned with other instances where there is a material change to the business, operations or affairs of a fund (e.g. change of fundamental investment objective, merger, etc.).

To address the comments raised regarding the regulatory burden, the Proposed Methodology requires the frequency of determining the investment risk level of a mutual fund to be at least annually, and within 60 days of the date of the Fund Facts or ETF Facts. This is a minimum frequency requirement and the investment risk level of the mutual fund should be reassessed more frequently, as appropriate.

Keeping the criteria outlined in the introduction above in mind, would you propose a different set of parameters or different frequency for monitoring risk rating changes?

If yes, please explain your reasoning. Please supplement your recommendations with data/analysis wherever possible.

Some of these commenters were concerned with how the proposed monthly monitoring process would apply to "borderline" mutual funds that sit on the higher end of a risk band range. These mutual funds would typically fluctuate between two risk bands from month to month, which, under the 2013 Proposal, would require more frequent re-classification. Where a fund manager is required to re-classify a borderline mutual fund more frequently, an amended Fund Facts and press release must be filed within 10 days of the last monthly calculation of the fund's SD. This is costly, burdensome and would likely lead to investor confusion.

One commenter commented that a risk classification methodology should provide a means to ensure that short-term fluctuations in investment risk levels are minimized. The 2013 Proposal seeks to avoid such short-term fluctuations by providing two tests associated with the monthly calculation. However, the commenter found these tests to be a bit confusing and potentially contradictory. The commenter pointed to the CESR methodology as being more intuitive, with less potential to provide contradictory signals.

One of these commenters recommended that any changes to a mutual fund's investment risk level should be a required discussion point in the fund's MRFP under National Instrument 81-106 Investment Fund Continuous Disclosure (NI 81-106) for the period of the change.

Other commenters agreed with the proposed process for monitoring investment risk levels. One commenter added that the process appears reasonable given that the purpose of the monitoring is to promptly alert investors of a material change in a mutual fund's investment risk level.

One commenter acknowledged that although necessary, monitoring and changing investment risk levels is time consuming and costly and these costs may well be passed on to investors.

13. Is a 10 year record retention period too long?

The vast majority of commenters suggested that the CSA limit the data retention period to 7 years. These commenters referenced paragraph 11.6(1)(a) of National Instrument 31-103 Registration Requirements and Exemptions. Another commenter suggested that a 7 year data retention period would be consistent with the Mutual Fund Dealers Association of Canada (MFDA) rules on the retention of documents.

After considering the comments received, the CSA has removed the requirement to maintain records for a ten year period. The requirement in securities legislation to maintain records for a period of seven years from the date the record was created applies.

If yes, what period would you suggest instead and why?

Two commenters were of the view that a minimum of 10 year be prescribed as a record retention period.

14. Please comment on any transition issues that you think might arise as a result of risk classification changes that are likely to occur upon the initial application of the 2013 Proposal.

According to several commenters, the 2013 Proposal would cause significant disruption to dealers and investors due to a large number of mutual funds moving to higher risk classifications. This will create a burdensome process for the advisors as there will be a need to review thousands of accounts and meet with thousands of investors to ensure ongoing suitability. Similarly, another commenter added that advisors and clients will have to determine whether the client should sell an investment as a result of the investment risk level change, potentially incurring taxable gains or losses or selling at an inopportune time, and raising costs for investors.

In response to commenters' concerns regarding unnecessary disruption to the industry, including dealers, we are proposing to retain the current five band risk scale. The proposed risk bands in the Proposed Methodology are also consistent with the IFIC Methodology which should minimize transition issues as the IFIC Methodology is widely used in industry. As a result, we expect any impact of implementing the Proposed Methodology to be minimal for fund managers, dealers and investors. Overall, we believe that the benefits of improved comparability of investment risk levels across mutual funds are proportionate to the costs of implementing a CSA mandated methodology.

How would IFMs and dealers propose to minimize the impact of these issues?

According to one commenter, another issue is the amendments of related regulatory documents as a result of fund risk ratings changes within the 10 day material change filing window. Fund managers may also be required to issue a press release to this effect. The commenter encouraged the CSA to consider the next filing of annual renewal of regulatory documents as a window for implementation of a risk rating change.

We are proposing that the Proposed Methodology be in-force after ministerial approval, i.e. 3 months after final publication of the proposed amendments. Once the Proposed Methodology is in force, mutual funds would be required to use the Proposed Methodology for each filing of the Fund Facts or ETF Facts, as applicable. This will allow mutual funds to transition to the Proposed Methodology according to their renewal prospectus schedule.

Commenters suggested various timelines for transition for both fund managers and for dealers. Commenters suggestions ranged from 6 -- 18 month transition timelines for fund managers to transition to the new risk classification methodology, followed by 12 -- 24 months for dealers to adjust and respond to the risk classification changes arising from implementation of the 2013 Proposal.

One commenter told us that a two year transition period should be sufficient for implementation, in recognition of the annual cycle followed by most fund managers in updating Fund Facts, i.e. by the end of two years after the requirement taking effect, all updates will have been completed.

In terms of the potential impact to dealers, advisors and investors, three commenters suggested that the CSA work closely with the self-regulatory organizations (SROs) to determine a suitable time period to allow dealers and advisors to consider the impact on investors of holding a mutual fund that has an investment risk level change as a result of the transition to the 2013 Proposal. In addition, the CSA and/or SROs should advise that a change in the assigned investment risk level from the adoption of the 2013 Proposal does not mean that the investment risk level of the fund has changed. Furthermore, investors should not necessarily be redeemed out of the particular fund due solely to the implementation of a mandated methodology. Commenters recommended that SROs publish guidance alongside proposed consequential rule changes so that the stakeholders can provide timely input to both the CSA and the SROs on the proposed means to achieve the stated regulatory objectives.

The CSA will continue to keep the SROs engaged as we proceed with implementation of the Proposed Methodology.

One commenter suggested that when developing transition to any new rules, it is of utmost importance that the CSA keep in mind: (i) the ongoing work within the industry to comply with the Client Relationship Model -- Phase 2 (CRM2) requirements that came into force in July 2013 and that any changes to investment risk levels of mutual funds can only be put in place at the earliest towards the end of 2016 or the beginning of 2017; and (ii) the recent choice of the CSA of mid-month dates, such as May 13 and June 13 (Fund Facts) and July 15 (CRM2), has significant implications for industry participants and the commenter urged the CSA to return to using calendar month-end dates, as well as dates that have a logical linkage to the new requirements and common industry timing, in order to ease transition. Finally, any changes in risk classification should also be communicated to existing investors, perhaps by reference in the semi-annual and annual MRFPs required by NI 81-106.

The CSA are mindful that there are 2 concurrent workstreams relating to the Proposed Methodology and the ETF Facts. We will endeavour to co-ordinate transition periods for final amendments where possible.

To reduce the costs and logistical complexity to fund managers resulting from successive, incremental changes to form requirements, the commenter strongly encouraged the CSA to, where possible, consider aggregating proposed changes through the use of transitional periods such that they apply at the same time.

Fund mergers/ conversions

A few commenters suggested that the 2013 Proposal should provide specific guidance around how to determine the investment risk level of the continuing fund in the case of a fund merger.

The Proposed Methodology has been amended to include specific provisions where there are fundamental changes to the investment objectives of a mutual fund or a reorganization or transfer of assets of a mutual fund.

Another commenter felt that in a fund merger situation, there needs to be clear rules surrounding the use of historical returns, particularly if the mutual funds are from distinctly different asset classes or investment strategies. It may be beneficial to set a limit on how much the investment risk level on the newly merged investment fund can be lowered.

One commenter suggested that where an older fund's series of securities are being merged into a newer series of securities of the same fund, the returns of the older series of securities should be used to calculate the SD.

One of these commenters also wondered how to handle the situation where a closed-end fund converts to a mutual fund. The commenter wondered if the CSA will permit using historical closed-end fund data.