T.Rowe Price

T. ROWE PRICE ASSOCIATES, INC.

WWW.TROWEPRICE.COM

P.O. Box 89000
Baltimore, Maryland
21289

100 East Pratt Street
Baltimore, Maryland
21202·1009

Toll Free
Fax

Via Electronic Mail
January 13, 2016

800·638· 7890
410· 345-6575

Mr. Brent J. Fields, Secretary
U.S. Securities and Exchange Commission
100 F Street, NE
Washington, DC 20549-1090

Re:
Open-End Fund Liquidity Risk Management Programs; Swing Pricing;
Re-Opening of Comment Period for Investment Company Reporting
Modernization Release
File Nos. S7-16-15; 87-08-15
Dear Mr. Fields:
T. Rowe Price Associates, Inc. ("T. Rowe Price"), as sponsor and investment adviser to
the T. Rowe Price family of mutual funds ("Price Funds"), appreciates the opportunity to
comment on the above-referenced proposal (the " Proposal") issued by the Securities and
Exchange Commission ("SEC"). As of September 30, 2015 , T. Rowe Price and its affiliates
managed approximately $725.5 billion in assets, and the Price Funds comprised 179 funds with
aggregate assets of approximately $466.0 billion.
Overall, we are supportive of the Proposal's requirement for mutual fonds to adopt and
maintain a liquidity risk management program. We believe it will help the SEC, as the primary
regulator of the asset management industry, to more effectively monitor and oversee the
activities of mutual funds. While we are in general agreement with the underlying principles of
the Proposal, we offer below some specific recommendations that we think will further enhance
its effectiveness in light of our serious concerns with the Proposal ' s liquidity classification
requirements. We note our general concurrence with the views expressed by the Investment
Company Institute ("'ICI") and the Securities Industry and Financial Markets Association
("SIFMA") in their comment letters and encourage the SEC to consider the additional issues we
raise and an alternative liquidity classification framework that is based on SIFMA's approach.
The following summarizes our perspectives with respect to the Proposal:

Liquidity Risk Management Programs
•

Risk-Based Approach: We agree that funds should be required to maintain
liquidity risk management programs operated by risk management professionals

1

'L

T.Rowel\ice

INVEST WITH CONFIDE NCE

and subject to oversight by fund boards, and we also support the proposed
codification of the 15% standard asset limit.

•

Proposal's Liquidity Classification Categories: We have concerns that the
Proposal's specific liquidity classification requirements will create a complex
liquidity compliance regime that will likely prove to be of little benefit to
practicing fund managers, regulators, or the public. We believe the proposed
requirement for funds to classify the liquidity of portfolio assets into six specified
"convertible to cash" categories requires a degree of subjectivity that will make
the classifications less relevant and useful as an indicator of fund liquidity. In the
body of this letter, we recommend an alternative approach that is based off of the
alternative described in SIFMA's letter and more in line with how liquidity is
determined, analyzed and managed by funds. While we think the SEC ' s approach
is well-meaning, we believe its level of preciseness is illusory and, in reality,
wmecessary for funds to effectively manage thei r liquidity.

•

"Convertible to Cash at a Price that does not Materially Affect the Value"
Standard: Both the liquidity classification framework and the three-day liquid
asset minimum require funds to assess the relative liquidity of each portfolio
position based on the "number of days within which it is determined, using
information obtained after reasonable inquiry, that the fund' s position in an asset
(or a portion of that asset) would be convertible to cash at a price that does not
materially affect the value of that asset immediately prior to sale." The concept of
converting a holding to cash "at a price that does not materially affect the value of
that asset immediately prior to sale" is highly subjective, and will be burdensome
and costly to implement in practice, especially for funds that could hold hundreds
of portfolio positions. We recommend instead that funds primarily assess liquidity
for purposes of classifying assets into liquidity buckets with reference to the time
it would take to sell or liquidate a position without considering market impact.

•

Liquidity Classifications - Need for Safe Harbor: If the SEC adopts the rule
as proposed, in light of the high degree of subjectivity involved in classifying the
liquidity of portfolio assets, we believe the SEC should more explicitly provide
that funds and their affiliates will not face liability for errors in classification
unless the error is material under normal conditions and the fund or affiliate acted
knowingly or recklessly.

•

Alternative to the Three-Day Liquid Asset Minimum: We recommend, as an
alternative to requiring a three-day liquid asset minimum, that the SEC require
funds to determine whether to establish a highly liquid asset percentage (as
defined further in the letter), considering all of the liquidity tools available to a
fund.

•

Disclosure and Reporting: If the rule is adopted as proposed, we suggest that
the SEC require monthly reporting on Form N-PORT of the percent of a fund's
assets in each of the classification categories to the SEC but that the information
2

ea.

T.Roweltice

INV E S T WIT H C ONF I DEN CE

should not be made public. We note that our comment above concerning
protection from liability resulting from errors in classification is particularly
applicable to the extent that subjective, asset-by-asset classification information is
publicly disclosed. In addition, we oppose the requirement for a fund to file a line
of credit agreement as an exhibit to its registration statement.

•

Board Approval of Liquidity Program: We are supportive of fund boards
providing oversight of funds' liquidity risk management programs but we do not
believe that fund boards should be required to approve specific elements or
components of funds' Liquidity Programs, including a fund's three-day liquid
asset minimum. We recommend that the SEC look to the fund compliance rule,
Rule 38a-1, as a model for framing the board 's oversight function over liquidity
risk management.

•

Factors for Consideration - Relationship of an Asset to Another Portfolio
Asset and the Fund's Use of Derivatives: The factors for both the liquidity
classifications and the 3-day liquid asset minimum suggest that a fund should treat
assets used by the fund to "cover" derivatives transactions as having the same
liquidity classification of the derivative instruments they are covering. Also, with
respect to the liquidity classifications, in situations where a fund purchases a more
liquid asset in connection with a less liquid asset and the fund plans to transact in
the more liquid asset only in connection with the less liquid asset (i.e. ,
"hedging"), then both assets should be classified in accordance with the less
liquid asset. We request the SEC remove these factors from the final rule.

•

Compliance Period: Based on the fact that we manage a large number of funds
that hold a broad range of securities, we request a compliance date of 30 months
after the effective date to comply with the Proposal rather than the proposed 18
months.

Swing Pricing

•

Operational Challenges for U.S. Funds: We support the SEC's goal of
protecting investors by reducing potential dilution resulting from high volume
fund flows, and we also support the SEC's proposal of providing funds with the
option to use swing pricing as a potential tool for achieving that goal. We note,
however, that given the nature of fund distribution channels, funds often do not
have access to timely information on net inflows and outflows and, therefore, will
have difficulty accurately assessing when a fund's swing threshold has been
exceeded. We would like the SEC, before adopting a final swing pricing rule, to
consider ways it can require intermediaries to provide more transparency so funds
can have intraday access to fund flow information.

•

Swing Factor Determination - Market Impact Costs: We believe that mutual
funds should be permitted but not required to consider market impact costs when

3

I.

T.Roweltice

INVEST WITH CONFIDENCE

determining their swing factor. Determining these costs involves subjective
predictions that significantly add to a Jack of precision in the swing facto rs.

•

Compliance Period: We believe the SEC should provide for a compliance date
that is one year after the effective date.

Our detailed comments on these issues are set forth below.

I.

Liquidity Risk Management Programs
Risk-Based Approach

We support a rule requiring funds to adopt written liquidity risk management programs
("Liquidity Programs") that are risk-oriented and principles-based. We also support the
proposed requirement to codify existing SEC guidance on illiquid assets, that a fund not acquire
any asset that may not be sold or disposed of in the ordinary course of business within seven
calendar days at approximately the value ascribed to it by the fund ("15% standard asset") if,
immediately after the acquisition, the fund would have invested more than 15% of its assets in
15% standard assets. However, we believe that certain features of the proposed Liquidity
Programs, particularly the proposed liquidity classification system, should be modified.

Proposed Relative Liquidity Classification Categories
In general, we support the proposed requirements that funds assess, periodically review,
and manage their liquidity risk. However, we believe the proposed requirement for funds to
classify the liquidity of portfolio assets (and portions of portfolio assets) into six specified
"convertible to cash" categories is at the same time overly prescriptive and subjective, and that if
it were to be adopted as proposed, funds would need more guidance from the SEC on how to
classify specific asset types.

Overall, the proposed Liquidity Program is intended to reduce the risk that funds will be
unable to timely meet their redemption obligations pursuant to section 22(e) of the 1940 Act and
other regulatory requirements, diminish potential investor dilution, and allow funds to more
effectively manage liquidity risk. These requirements are intended to protect investors and
enhance the fair and orderly operation of the markets. The SEC believes that the proposed sixbucket classification system is more responsive to variation in the liquidity of funds' portfolio
positions than the current framework in which funds classify a position as either "liquid" or
unable to be sold or disposed of within seven days at approximately its stated vaJue ("illiquid").
In addition, under the proposal, funds would be required to report the liquidity classification of
each po1tfolio position as part of each fund's monthly portfolio holdings reporting on proposed
Form N-PORT. The SEC believes that this data would assist the SEC in assessing risks and
trends with respect to funds' portfolio liquidity and would aJso allow investors to better
understand and gauge a fund's overall liquidity risk profile.
In practice, however, we believe the proposed six-bucket classification system would be
operationally difficult to implement, of little benefit to fund investors, and an ineffective
4

T.Roweftice~

IN VEST WITH CONFIDENCE

liquidity tool for fund managers. The six-bucket classification system is highly subjective and
requires fw1d advisers to classify every fund holding (or portions of a holding) into one of six
specific, time-based buckets based on an estimation of market impact that cannot be exact.
Because of the high degree of subjectivity involved, we believe the Proposal's bucketing
approach will not be applied consistently across the fund industry and funds could classify the
same holding differently. This will create liability concerns for fund managers, who, under the
proposed rnle, will be required to publicly disseminate these classifications in Form N-PORT
quarterly. More importantly, it will make comparisons across funds less relevant for investors
and regulators, and would likely create misleading comparisons and outcomes.
The SEC recognizes in the Proposal that an asset's liquidity is often fluid and dynamic
and attempts to provide fund advisers with the flexibility to react to changing market conditions
by applying specific factors to assess and classify each position in the po11folio. However, we
believe that this approach, coupled with the degree of precision required to classify assets in the
way the SEC has prescribed, allows for too much interpretation and subjectivity, especially for
fixed i11come instruments and other instruments traded over-the-counter ("OTC"). Since there
are no generally accepted methodologjes for making the liquidity determinations that the SEC
has proposed, there will inevitably be a wide variety of liquidity assessments for the same or
similar assets, especially since the assessment of how long it will take to liquidate a particular
position could :fluctuate over time depending on market conditions and each fund manager's
assessment of market impact.
Further, the asset classification categories are represented by compressed time intervals,
which will be difficult if not impossible to predict with any precision or confidence. The trading
determination of whether a holding can be sold in three business days as opposed to four
calendar days without materially moving the market price can be made with little confidence,
much less with enough confidence to report the outcome of this judgment in regulatory filings
that are disclosed to the public. Further, each classification would be publically reported in Form
N-PORT in a format that appears objective and suggests that investors should use the data for
comparability across funds when in fact the classifications are subjective and imprecise.
Contrary to the SEC's intent, the data regarding liquidity classifications will not provide the
public with materially useful information about a fund's liquidity as the classifications will
greatly vary across fund complexes and may appear to be arbitrary even though such variances
would be related to bona fide differences in judgment.
Because the proposed classification scheme involves such a great deal of subjectivity,
funds that are more conservative in classifying their portfolio positions (by treating their fund
holdings as requiring more days to convert to cash), will actually appear to be less liquid and
thus potentially higher risk. An adviser will have to make judgments as to the bucket in which a
security should be placed where there is flexibility in settlement periods or where a fund holds a
relatively large position with limited trading volume. For example, for a foreign security or OTC
security that could settle in three business days but might settle in four calendar days based on
market convention, more conservative funds may bucket the position as convertible to cash
within 4-7 calendar days while less conservative funds would classify the asset as conve1tible to
cash within 2-3 business days. This would make the less conservative fund appear less risky and
more liquid to investors and other third parties even though the funds hold the same security. It
5

T.Roweltice~

IN VEST WITH C ON F l DEN C E

would also allow the less conservative fund to count an asset that other funds might consider less
liquid as a three-day liquid asset for purposes of the three-day liquid asset minimum.
Moreover, because the liquidity classification framework places emphasis on the size of a
particular position, a large fund will be placed at a disadvantage and will likely look less liquid
and more risky than a smaller fund with the exact same weightings of portfolio positions simply
because the larger fund has a larger position in the same security. For example, a very large,
diversified lar~e-cap U.S. equity fund, typically considered highly liquid at both the asset and
portfolio level , may not appear that way when compared to a smaller fund with the same
portfolio makeup as both funds could reach different conclusions as to how to categorize their
assets. We note that very large funds can typically handle a greater dollar volume of flows than
smaller funds with the same portfolio positions. The Division of Economic and Risk Analysis
study ("DERA Study"), for example, found that flow volatility decreases and portfolio liquidity
increases as fund size increases. The proposed classification system, however, does not take this
into consideration. Even if a larger fund and a smaller fund with the exact same portfolio
composition use the same methodology to bucket their assets, the larger fund will necessarily
look less liquid since it holds larger positions that may be spread across multiple buckets that
show longer time periods to liquidate.
In addition, the Proposal does not provide an adequate measure of liquidity for all asset
types. While the factors in the Proposal may translate well with respect to most types of equity
exchange-traded securities, the data required by many of the factors is not readily available for
fixed income securities, especially for fixed income securities of issuers in foreign or emerging
markets. Certain fixed income securities frequently trade based on an issuer's overall debt
structure and rating, not by the individual CUSIP. The liquidity characteristics of certain other
fixed income securities can be determined more by the homogenous nature of those securities
rather than individual characteristics of a particuJar security. The market factors that drive asset
class liquidity tend to be heightened in times of market uncertainty or stress, thus further
reducing the value, if any, of a CUSIP-by-CUSIP liquidity analysis. Moreover, infrequent
trading of certain fixed income securities can lead to funds using stale and inaccurate liquidity
measures in the calculation of a portfolio 's liquidity. Many municipal securities, for example, are
highly liquid, and smaller volumes can typically be sold in one to three business days. The
proposed factors, however, require funds to consider the frequency of trades or quotes in
assessing a position's liquidity. There are roughly 1.5 million different municipal CUSIPs in the
market, and the daily trading volume for these securities is typically about plus or minus l % of
the outstanding municipal CUSIPs. Because of this low daily trading volume, many municipal
securities could be classified in the more illiquid buckets under the SEC's proposed classification
system based on its factors and subjective market impact analysis when in actuali ty they are
typically highly liquid.
We note that the SEC suggested in the Proposal that external vendors will be able to
supply the necessary liquidity classifications, but we do not believe that this is a realistic
1

Paul Hanouna, Jon Novak, T im Riley & Christof Stahel, Division of Economic & Risk Analysis, SEC, Liquidity
and Flows of U.S. Mutual Funds 3 (Sept. 2015), available at https://www.sec.gov/dera/staff-papers/whitepapers/Iiquidity-white-paper-09-2015 .pdf.

6

Ii.

T.Roweltice

INVEST W IT H CONF I DENCE

solution. The proposed liquidity classifications lack objectivity, and third-party vendors are even
fu1ther removed from the trading determinations required in order to assess market impact on a
fond's portfolio ho ldings. We are not convinced that third-party vendors will be able to develop
the tools to address all of the proposed factors required to classify each portfolio position's
liquidity. Furthermore, even if such services were developed by third-party vendors to meet the
Proposal 's requirements, we would have reservations about subscribing for such data in view of
the level of subjectivity involved.
"Convertible to Cash at a Price that does not Materially Affect the Value" Standard

Both the liquidity classification framework and the three-day liquid asset minimum
require funds to assess the relative liquidity of each portfolio position based on the "number of
days within which it is determined, using information obtained after reasonable inquiry, that the
fund's position in an asset (or a portion of that asset) would be convertible to cash at a price that
does not materially affect the value of that asset immediately prior to sale." Assessing how long
it will take to convert an asset to cash without "materially affecting the value" of that holding
before its sale may be impossible to assess with any reasonable accuracy, particularly with the
time precision and extended time horizons proposed by the SEC. We recommend instead that
funds primarily assess liquidity with reference to the language that the SEC adopted in Form PF:
"assuming no fire-sale discounting." We also recommend incorporating the concept of "good
faith estimates" and basing such estimates on current market conditions.
Under the Proposal's standard, funds and fund advisers will have to make significant
judgments about the how quickly an asset can be sold at a pa11icular price prior to actually
selling the asset, and this cannot be accomplished with any precision or confidence, particularly
for municipal bonds and other OTC instruments with limited trading volumes. An asset's price
will change in response to a variety of forces in the market, and it is extremely difficult to
distinguish the impact of a fund's transactions from other market forces, especially prior to
executing the trade.
Additionally, the SEC shouJd not require funds to reference the amount of time it would
take to convert an asset to cash in all of the buckets. We note that there are many instances in
which the settlement time of a particular transaction may be significantly further in the future
than sale time. Our funds have successfully managed these timing differences in settlement
periods through their portfolio structure and asset mix; the day-to-day buy, sell and hold
decisions made by our investment professionals; and the use of other liquidity tools that the SEC
references in the Proposal. Under the SEC's proposed standard, assets that have longer
settlement times may make a fund look significantly less liquid, when in reality, the fund can
easily manage these disparate settlement periods. Instead, funds should make a good faith
estimate for liquidity based on current market conditions and assuming no fire-sale discounting.
T. Rowe Price's Approach

Our liquidity risk management framework as it cunently exists and as it is evolving
provides portfolio managers, investment division leadership, and risk personnel with insight into
the T. Rowe Price portfolios' expected ability to satisfy potential client withdrawals. The
7

'L

T.Roweltice

INVEST WITH CONFIDENCE

framework is intended to apply broadly across the firm, but is constructed to take into
consideration differences for various asset classes and security types. The risk management
framework is administered and overseen by risk professionals at T. Rowe Price but relies heavily
on the advice of subject matter experts from the investment divisions and trading desks. It aligns
with how our portfolio managers consider liquidity in their investment management decisionmaking process because it is a holistic assessment of the po11folio that involves a security
classification utilizing liquidity-related data where readily available (for example, daily trading
volume for exchange-traded equity securities) but also more qualitative information where such
data is not obtainable or not as relevant. During our review of a portfolio's liquidity, we make
certain assumptions for eacb asset based on the characteristics and available data and other
liquidity-related information for each instrument's asset class. For example, we assume that all
investment-grade bonds have a similar liquidity profile. We do not drill down to look at, for
example, each investment-grade bond' s issuer, maturity, and issue date for purposes of
classifying its liquidity. The Proposal is a significant departure from our approach because it
requires a quantitative analysis that not only looks at each particular security but requires
advisers to analyze each position 's (and portions of a position) particular characteristics
separately.
Liquidity reporting primarily focuses on assessing the ability to meet one-day
redemptions equal to the 99th percentile (gross redemptions) experienced by a portfolio during a
"normal" market environment. The reporting also provides similar information assuming a
·'stressed" envirorunent, which is based on the portfolio's ability to meet a redemption that is two
times the size of the 99th percentile (gross redemptions) and assumes reduced levels of market
liqui dity. While the fund board receives regular presentations from the chief risk officer and the
risk team on liquidity risk management, unlike the Proposal, the board is not responsible for
determining or approving any liquidity risk metrics as those determinations would be more
appropriately made by risk professionaJs with the input of both equity and fixed income
investment professionals who are subject-matter experts in the securities they cover and trade.

Costs of the SEC's Liquidity Classification System
Building and maintaining the proposed six-bucket classification system would pose a
significant burden on T. Rowe Price, with fund shareholders ultimately carrying a majority of the
expense. All securities (and portions of such positions) held within a fund complex would need
to be analyzed, assessed and then assigned to one of the six liquidity buckets. An individual fund
may have hundreds if not thousands of portfolio positions, and the T. Rowe Price family of
mutual funds held just under 44,000 holdings combined, as of December 31 , 2015. As
mentioned previously, our approach differs quite significantly from the six-bucket system
mandated by the Proposal, so we would have to significantly augment our system and develop
new processes to meet its requirements. We note that some firms, including T. Rowe Price, have
systems in place to compile some of the data required under the Proposal, but additional data will
have to be collected, managed and analyzed, particularly in order to capture the anticipated
market impact in selling a position. In addition, OTC instruments that do not have readily
available trading volume data, or whose trading volume is relatively low, would require
additional research and manuaJ intervention by trading professionaJs in order to assess all the
factors required to classify the security. Furthermore, under the Proposal, the majority of large
8

fi.

T.Rowelti.ce

INVEST WITH CONFIDENC E

positions would have to be assessed and potentially divided into several different buckets, and
we would need to build a process to be able to capture this element of the proposed bucketing
scheme.
Further, compiling and reviewing the data for our current holdings and newly purchased
assets would pose a significant burden, and the requirement for "ongoing review" of each asset
would place an even greater burden, on funds. At a minimum, funds would have to review all
liquidity classifications at least monthly in connection with Form N-PORT filings, and, as
described in the Proposal, funds holding assets whose liquidity depends considerably on current
market conditions would need to review their assets' liquidity classifications relatively often (the
Proposal suggests that such review could occur daily, or even hourly, depending on the
circumstances). This would be an extremely resource-intensive endeavor - the costs of which
would ultimately be borne by fund shareholders. While we have not attempted to try to estimate
the costs of building the SEC's classification system (and we know of no similar system or
service that is commercially available), we think the effort involved would be comparable to the
daily p1icing of a fund's portfolio, which could range in the millions of dollars for a fund
complex the size of T. Rowe Price. For many funds (for example, funds whose positions do not
have readily available quantitative data related to liquidity), the exercise of continuously
reviewing every position in every portfolio (which, as mentioned above, could be as often as
daily or even hourly) would be operationally similar to continuously fair valuing the fund's less
liquid portfolio positions. We question whether the potential benefits of the Proposal's
classification system would outweigh these costs.
Proposed Alternative:
We propose that the SEC adopt a framework that utilizes four classification categories
instead of six, with the factors outlined by the SEC in the Proposal considered guidelines rather
than codified in the rule text. This alternative is similar to the alternative proposed by SIFMA in
their comment response letter. Like the SIFMA alternative, the buckets would be based on a
spectrum from highly liquid (Category 1) to illiquid (Category 4). The factors and methodology
for allocating portfolio assets among the categories would differ among the categories, reflecting
the different levels of precision that are appropriate for different points along the spectrum, and
the purpose of each category.
Under this alternative, the adopting release would continue to make clear that funds may
consider additional factors, and that a fund would not be required to consider any particular
factor in classifying indi victual assets. In addition, the adopting release would make clear that a
fund would be permitted to weigh any one factor or group of factors over others, and could
ignore certain factors entirely, as it sees fit in categorizing a particular asset or asset class.
We note, for example, that with respect to exchange-traded equity instruments, funds can
use historical quantitative measures, such as a percentage of an issue's daily average tracling
volume ("DATY") (e.g., 20% of DATY), to determine an asset's liquidity classification. With
respect to many other securities traded in dealer markets or in developing markets, however,
there can be limited data available, and funds would be required to rely on more qualitative data.

9

T.Roweltice~

INVEST WITH CONF I DENCE

We note that if more trading volume and other liquidity-related data becomes available for such
asset classes, funds may be able to utilize a more objective bucketing approach for these assets.
In addition, under our proposed alternative, a fund would be pem1itted to classify an
entire asset class that the adviser believes is generally homogeneous (e.g., large-cap equity) and
would not be required to evaluate the liquidity of any particular asset in that asset class unless the
adviser knows or has a reason to know that the asset (or portion thereof) should fall within
another category. In such situations, advisers could have the option of classifying a position into
a different category and could disclose the exception on Form N-PORT.
At one end of the spectrum, Category 1 would correspond in most respects to the
Proposal's three-day liquid asset designation (i.e. , highly liquid assets that are convertible to cash
within three business days), with the differences noted below, and thus would include the type of
assets that would be expected to qualify for the "three-day liquid asset minimum" in the
Proposal. Assets in this category could be used towards a fund's "highly liquid asset
percentage," which we propose below as an alternative to the proposed "three-day liquid asset
minimum.'' At the other end of the spectrum, Category 4 would include assets currently
considered illiquid under the "1 5% guidance," and that would be considered ·' 15% standard
assets" tmder the Proposal.
The second and third categories would be used to classify assets that fall between the two
ends of the spectrum. For many funds , these middle categories will cover the great majority of
fund assets, and thus the bulk of the assets that portfolio managers rely on to implement their
investment strategies and that they will purchase and sell for that purpose. In these middle
categories, fund assets are considered liquid, but are not relied on as "highly liquid," and can be
viewed as more or less relatively liquid depending on a variety of qualitative (subjective) and
quantitative (objective) factors.
We note that some of our greatest concerns with the Proposal arise from the following
aspects of the proposed classification system: (1) identifying a precise number of "days to cash"
for all buckets; (2) applying the standard "without materially affecting" the asset's price; and (3)
classifying each "portion" of every asset separately. We incorporate the concept of "days to
cash" in Category 1, where we believe it can be workable when applied in the appropriate
context and purpose.
Category 1 is designed to identify highly liquid assets, both for general liquidity risk
management purposes and for purposes of our recommended highly liquid asset percentage
aJternative, described below. Because of this purpose, it is appropriate to consider a stated
number of days (in this case, approximately three business days) expected to convert the position
to cash in the context of normal trading (based on normal trading lots). An asset's classification
in Category 4 would be based on expected sale within seven days at approximately its stated
value, in accordance with the longstanding 15% guidance. Classification in Categories 2 and 3
would be based on a variety of qualitative and quantitative considerations. Other than with
respect to the three-day and seven-day standards for Categories 1 and 4, the timing of asset sales
would be a factor but would not be identified with reference to specific range of days.

10

'L

T.Roweftice

INVEST WITH CONFIDENCE

Classification of assets in all categories would represent a good faith estimate of liquidity
based on current market conditions. As discussed above, classification would not require an
evaluation of whether the asset can be sold "without materially affecting" the asset's price
immediately prior to sale. We note that if the SEC believes a boundary must be part of the rule,
we recommend instead requiring an assumption that the asset would be sold with no fire-sale
discounting. This boundary is based on Fonn PF, Question 32, which states in part: "Use good
faith estimates for liquidity based on market conditions over the reporting period and assuming
no fire-sale discounting." Unlike Form PF, however, we recommend an estimate based on
current market conditions at the end of the reporting period, rather than over the reporting period.
Lastly, with respect to the proposed treatment of position size (reflected in the Proposal
as a requirement that classifications be assigned to "portions" of a position), our recommended
alternative would permit but not require funds to separately classify portions of a single holding.
Moreover, as discussed above, we believe that the position size factor would most significantly
impact large funds , which are more likely to have holdings that represent a significant portion of
the market but typically have greater overall liquidity and resources to manage liquidity
effectively.
The following is a description of the four liquidity categories we propose:

•

Category I. Cash, cash equivalents, and assets that the fund believes it can liquidate (sell
and receive settlement proceeds) within approximately tlu-ee business days in the context of
n01mal trading (also referred to as "highly liquid" assets).

•

Category 2. Assets that are considered liquid in ordinary markets, but which may become
less liquid in stressed conditions or may, for other reasons, be expected to require more than
three business days to liquidate.

•

Category 3. Assets that are considered less liquid than assets in Category 2 (e.g., there are
fewer active participants in the relevant market and execution is more sporadic and in smaller
sizes), but are still considered liquid (and thus not appropriate for classification in Category
4).

•

Category 4. Assets that the fund believes could not reasonably be expected to be sold by the
fund within seven calendar days at approximately the value ascribed by the fund (i.e. , " 15%
standard assets," as that term is used in the Proposal, which we refer to as "illiquid").

We believe that this alternative offers many advantages to the six-bucketing approach
described in the Proposal. First, the classification system is based on four buckets that encompass
a larger range than the proposed six buckets, which are far too granular and precise. The use of
fewer buckets wou1d provide fund managers with more flexibility in bucketing assets. A
potential benefit of a more flexible, less precise bucketing scheme is that it reduces some of the
liability concerns discussed in this letter. In addition, fewer buckets may also reduce potential
shareholder confusion because there will be less room for variation between funds. This
alternative would also allow fund managers to group homogeneous assets together, which would
be considerably less burdensome and would provide a significant cost savings.
11

T.Roweltice~

INVEST WITH CONF I DENCE

We recognize and agree with the SEC on the importance of being able to convert some
assets to cash within tllJee business days to meet redemption requests. We believe that Category
1 as described above appropriately requires funds to consider which assets can be used to meet
shareholder redemptions. However, using this standard for all assets, including the ones we know
cannot be converted to cash within three business days, is unnecessarily burdensome.
Consequently, for aJl categories other than Category I, the classifications would be made using
a good faith estimate for liquidity based on current market conditions and assuming no fire-sale
discounting.
We note that even though the proposed alternative reduces some of the subjectivity
involved in classifying assets, it still allows for variation between funds. Consequently, in the
adopting release, we ask that the SEC provide specific guidance on how to classify different
asset types with a variety of examples of how the factors should be applied.
Liquidity Classification - Need for Safe Harbor

We believe that the subjectivity of the proposed liquidity classifications raises significant
liability concerns for funds and their advisers. As proposed, each portfolio position's liquidity
classification will be reported to the SEC monthly and disclosed quarterly on Form N-PORT.
Section 34(b) of the 1940 Act provides that it shall be unlawful for any person to make any
untrue statement of a material fact in any registration statement, repo1t, record, or other
document filed or transmitted pursuant to the 1940 Act or the keeping of which is required
pursuant to Section 3 l(a) of the 1940 Act. The liquidity classifications reported on Form NPORT will be subject to liability under Section 34(b), and it is also possible that information
concerning liquidity classifications could be held to be information, the keeping of which is
required pursuant to Section 3l(a) and, thus, subject to liability under Section 34(b). We note
that Section 34(b) does not requfre a violator to act willfully, and the SEC has ruled that intent or
knowledge is not required to establish a violation.
As discussed above, however, if adopted as proposed, liquidity classifications will be
highly subjective, with the potential for funds classifying the same or similar positions very
differently. As mentioned above, we believe that these liability concerns will be greatly reduced
if the SEC pursues an approach similar to the alternative we suggest and provides more specific
guidance on how to apply the classification scheme to specific asset types. However, in any case,
we believe that the SEC should affirmatively provide a safe harbor so that liability will not attach
for errors in a position 's liquidity classification unless the error is material under normal market
conditions and the fund or other person responsible for the filing acted knowingly or recklessly.
Alternative to the Three-Day Liquid Asset Minimum

We strongly oppose a three-day liquid asset minimum that would be approved by a
fund 's board. Portfolio managers continually consider the amount of liquid assets they need to
maintain in a portfolio. We believe that it would be appropriate for a liquidity risk management
program to include a requirement for portfolio managers, in conjunction with risk personnel, to
consider and for boards to review the amount of highly liquid assets maintained in their
portfolios in order to meet redemptions. However, the portfolio manager should retain the
12

fi.

T.Roweltice

I NVEST WITH CONFIDENCE

ultimate decision on the percentage of highly liquid assets that is prudent for the fund as part of
managing the fund 's investment program, which may take into account some of the factors that
the SEC has outlined including, shareholder redemption activity, shareholder concentration,
availability of borrowing opportunities, and other relevant factors.
Disclosure and Reporting

If the rule is adopted as proposed, we ask the SEC to allow funds to report liquidity on an
aggregate basis per classification bucket instead of at the individual security level. This would
help address tl1e issues discussed above relating to subjectivity and the differences in
classifications across funds and the potential liability associated therewith. Even though our
proposed alternative also involves security classification, we have concerns if this information
were to be publicly disclosed. Individual holding liquidity classifications will not provide
meaningful information to investors as these liquidity determinations would be provided out of
context as structured data in Form N-PORT with no information for the investor as to how it fits
into the fund's risk profile as a whole.
Alternatively, we would support a regime whereby individual security level
classifications would only be available to the SEC and that the SEC would only make public
aggregate liquidity information by bucket. We note that the SEC would also have access to a
fund 's asset-by-asset classifications during an exam.
While we currently file a line of credit agreement as a material contract with respect to
ce1tain Price Funds as part of their registration statements, we believe the agreement itself offers
little value to investors since these agreements and their exhibits are complex and lengthy legal
documents with technical terms. In addition, filing these agreements on EDGAR is laborintensive because they are so long and not easily convertible into EDGAR formats . We further
note that certain census-type information regarding line of credit arrangements can be reported in
funds' Fonn N-CEN, including, for example, whether the fund has a committed line of credit,
the size of the line of credit, the name of the institution with which the fund has the line of credit,
whether the line of credit is shared among other funds , whether the line of credit was drawn on,
the average amount of credit used, and number of days the credit was outstanding. We believe
that this type of summary infonnation will be more useful to shareholders than parsing through
an entire line of credit agreement itself, and we therefore support its inclusion on Form N-CEN.
Lastly, we note that the SEC would still have access to a fund's line of credit documentation
during an exam. Therefore, we recommend eliminating this requirement from any final rule.
Board Approval of Liquidity Program

We agree with the SEC that independent oversight by a board over a fund' s Liquidity
Program is appropriate, and we note that such a requirement is similar to board approval of a
fund's compliance program required by Rule 38a-1 under the 1940 Act. Similar to a fund's Rule
38a-1 compliance program, however, we do not believe that fund boards should be required to
approve specific elements or components of funds' Liquidity Programs, including the three-day
liquid asset minimum. As noted above, establishing a three-day liquid asset minimum will
involve a fact-intensive, technical analysis of the multiple factors required by rule 22e-4. We
13

fi.

T.Roweltice

I NV EST WITH CONF I DENCE

believe that the investment staff in conjunction with independent risk and/or compliance
personnel, rather than the fund board, is most appropriate to make these determinations and that
a fund's board should primarily serve in an oversight role with respect to the process itself, as
opposed to the specific outcomes of the process.
Instead ofrequiring a fund board to approve a fund's three-day liquid asset minimum, we
suggest that the board oversee the process utilized by portfolio managers and risk personnel to
determine the percentage of liquid assets maintained in a portfolio. Fund boards will receive at
least annual reports regarding the adequacy and effectiveness of a fund's liquidity risk
management program, which should include a report on the operation of the liquid asset
percentage, including whether it was re-set at all during the preceding reporting period. This
approach is more consistent with the traditional role of the fund board, where the board is
responsible for monitoring operational areas that present material risks or conflicts of interest but
not required to make specific determinations about highly detailed, operational issues in which
they are not directly involved. Furthermore, funds will retain flexibility to quickly adjust their
liquid asset percentages in response to changing market conditions without having to call a
forma l fund board meeting. Lastly, we note that board approval of the liquid asset percentage is
not necessary for the protection of investors assuming the board retains oversight with respect to
the process.
We agree that a fund's investment adviser or risk officers administering the fund's
liquidity risk management program should be required to submit written reports to the fund's
board concerning the adequacy of the fund's liquidity risk management program, including the
fund 's liquid asset percentage, and the effectiveness of its implementation. However, we oppose
the proposed requirement that funds maintain a written record of the assessment of each of the
factors set forth in the proposed rule when setting and adjusting each fund 's liquid asset
percentage. We are not aware of any similar recordkeeping requirement under the 1940 Act for
other board determinations that involve a fund 's investment program. Instead, consistent with
our position that the fund board should serve in an oversight capacity, we feel that the factors
should be considered by the fund's adviser and, to the extent relevant, discussed more generally
in the written report submitted to the fund ' s board.
If the rule is adopted as proposed, given the fact-intensive and subjective nature of many
of the factors, we believe that the final rule should provide fund boards with a safe harbor in
approving specific elements of the liquidity risk management programs and clarify that funds
and/or boards are not required to consider all of the factors set out in the Proposal , particularly if
those factors are not applicable. The rationale for the board's safe harbor would be consistent
with the recommendation above with respect to a safe harbor for disclosure by the fund of its
liquidity classifications.
Factors for Consideration - Relationship of an Asset to Another Portfolio Asset and
the Fund's Use of Derivatives
As proposed, the factors for both the liquidity classifications and the three-day liquid
asset minimum require funds to treat assets used by the fund to "cover" derivatives and other
transactions by applying the liquidity classification of the derivative instruments they are
14

'L

T.Rowellice

IN V EST WITH CONF I DENCE

covering. We believe this approach is problematic and should be removed from the final rule. As
noted in the Proposal , Investment Company Act Release No. 10666 ("Release 10666") states that
"[a] segregated account freezes certain assets of the investment company and renders such assets
unavailable for sale or other disposition. " However, these segregated assets would only be
"frozen" and "unavailable for sale or other disposition" for the period that they are used to cover
a specific transaction. At any point, a portfolio manager may decide to unwind a specific asset
used as coverage and may replace that asset with another liquid asset. We note that this may
change very quickly as portfolio managers consider and re-consider the assets that are tied to a
transaction as coverage.
The proposal notes that, in accordance with Release 10666, segregated assets may be
replaced by other appropriate non-segregated assets of equal value, and when they are so
replaced, formerly segregated assets would no longer be considered unavailable for sale or other
disposition. However, as proposed, until those segregated assets are unwound, funds would still
be required to classify those assets by referencing the liquidity of underlying instrument. Since
segregated assets can be unwound quickly (for example, the following day), we believe that
treating assets used by the fund to "cover" derivatives and other transactions using the liquidity
of the derivative instruments they are covering provides an inaccurate representation of a fund's
liquidity and effectively counts an otherwise liquid asset as illiquid based on a regulatory
interpretation, while ignoring its true liquidity.
With respect to the factors to consider for liquidity classifications, the proposal states that
in situations where a fund purchases a more liquid asset in connection with a less liquid asset and
the fund plans to transact in the more liquid asset only in connection with the less liquid asset
(i.e., "hedging"), then both assets should be classified in accordance with the less liquid asset.
However, the proposal fai ls to recognize that hedging is a portfolio management decision. Even
if an instrument is originally intended to be used as a hedge for some other, less liquid asset in
the portfolio, it is not required to be used as such, and the portfolio manager may decide to
transact in the more liquid asset outside of any reference to the less liquid asset. Deciphering a
fund manager's intentions with respect to the relationship between portfolio positions introduces
a level of complexity that is unnecessary for purposes of a fund 's overall liquidity risk
management and will increase the costs of compliance. Consequently, for liquidity classification
purposes, the two assets should be treated separately.
In addition, we note that some strategies (for example, certain bond funds) utilize
derivatives to manage overall portfolio characteristics. For example, a short term bond fund may
purchase a futures contract as a means of managing the overall duration or interest rate
sensitivity of the portfolio. In such situations, the derivative is not related to any particular asset
in the portfolio. It is not clear how such a derivative's liquidity would be classified under the
proposed framework.
Lastly, we note that the proposal seems to suggest that derivatives are inherently more
risky and present greater liquidity risk than other, more traditional assets. We note that, in some
situations, derivatives may be more liquid than more traditional assets, including some bonds.
For example, many derivatives that reference bonds as w1derlying reference assets are exchange
traded and cash-settled, unlike the underlying bond, which may not be.
15

'I.

T.Roweltice

INVEST WI T H CONFIDENCE

Compliance Period
The SEC proposes a compliance date of 18 months after the effective date to comply with
proposed rule 22e-4, but smaller entities (groups of investment companies with net assets of less
than $1 billion as of the end of the most recent fiscal year) would have a compliance date that is
30 months after the effective date. As a large fund family, we believe that it will require more
than 18 months to develop internal processes and procedures to implement Liquidity Programs
for our funds, including the data and systems development work necessary to build a liquidity
classification system, and to seek board approvals, all in accordance with the requirements of the
Proposal. Larger fund complexes will face the same challenges and need as much or more time
as smaller fund families in complying with the Proposal. Accordingly, we request a compliance
date for all entities that is 30 months after the effective date.
II.

Optional Swing Pricing

Operational Impediments to Implementing Swing Pricing in the US.
We support the concept of swing pricing as an additional tool that open-end funds may
use to mitigate potential dilution by passing on purchase and redemption costs to the transacting
shareholders, rather than having those costs borne by remaining shareholders. We have found
the use of swing pricing to be an effective anti-dilution tool for our Luxembourg-domiciled
societe d'investissement a capital variable ("SICAV'') funds. However, we note that there are
serious operational challenges that will not make it possible to use swing pricing in the U.S. in
the same way that it is currently applied in other jurisdictions. For example, there is a 3-hour gap
between the cut-off time for orders received by our Luxembourg-based SICAV funds until the
time they strike their net asset value ("NA V''). During this time, the SICAV can collect
information on fund flows from their transfer agent in order to calculate whether any swing
thresholds have been breached and, if so, the swing factor that should be applied. No such gap
exists for our U.S. mutual funds. Our U.S. mutual funds produce their NA Vs as of 4:00 p.m. ET,
which is the same as the cut-off time for shareholder transactions.
Many investors in U.S. mutual funds do not purchase shares of a fund directly from the
fond itself but instead purchase and redeem shares from any one of several different types of
intermediaries, such as broker-dealers, fund platforms and retirement plans. Ultimately, all fund
purchase and redemption information is submitted to the fund's transfer agent, and the transfer
agent then keeps track of cash flowing into and out of the fund. However, intermediaries
typically net orders received from investors against each other and submit a single file containing
net or omnibus purchase and redemption information to the fund 's transfer agent. Thus, fw1ds do
not get a clear picture of total purchases and total redemptions but rather just net information.
Under rule 22c-2, purchase and redemption orders must be submitted by investors to
broker-dealers and other intennediaries by 4:00 pm in order to receive that day 's price, and,
pursuant to the SEC's rules, intermediaries are then allowed to provide the complete order
information to the fund transfer agent at a later time. Often, intermediaries do not provide this
information to the fund 's transfer agent until late at night or even the following morning.
Because funds do not have a complete understanding of fund flow information until after they
16

T.Roweltice~

INVEST WIT H CONF ID ENCE

strike their NAV, we are not sure it will be possible to accurately determine whether or not the
swing threshold was breached. Consequently, funds will be using imperfect and incomplete
information to adjust their NA Vs, as they will necessarily need to rely on interim information or
estimated flows from their intermediaries. There could be substantial costs and technology
hurdles in order for intermediaries to provide enough information prior to finalizing the
calculation of a fund's NAV to allow the fund to accurately determine whether a swing threshold
has been breached. Our experience with Rule 22c-2 tells us that it will not be easy to collect this
information from all intermediaries in a timely manner.
It is our understanding that this issue of material shareholder flow information not being
received until after NA V calculation is not present in other jurisdictions where swing pricing is
applied, as ownership of SICAV funds, for example, is heavily weighted towards direct,
institutional investors. This difference in shareholder base means that accurate flow information
can be received prior to NA V calculation, and in the case of SICAV funds, accurate flow
information is typically received three hours prior to NAV calculation. This timing allows for
swing pricing to operate with minimal risk of NA V errors due to misapplication of a swing
factor.

Given the potential for NA V errors if swing pricing is implemented with the current
timing of fund flow information, we strongly encourage the SEC to consider what changes are
necessary to its regulatory framework to require (or otherwise provide funds with the ability to
influence) intermediaries to provide accurate estimates of purchase and redemption information
prior to funds striking their NA Vs so that swing pricing can be an effective tool to mitigate
potential dilution. 2
Swing Factor Determination

We believe that mutual funds should be permitted but not required to consider market
impact costs when determining their swing factor. These costs are not easily quantifiable and
funds wilJ not be able to generate an accurate estimate of market impact costs.
Under proposed rule 22c-1 (a)(3), in determining its swing threshold, a fund would be
required to consider, among other things, any near-term costs that are expected to be incurred by
the fund as a result of net purchases or net redemptions occurring on the day the swing factor is
used to adjust the fund's NAV per share, including "market impact costs." Market impact costs
are defined in the Proposal as costs that are incurred when the price of a security changes as a
result of the effort to purchase or sell the security. The SEC recognized in its Proposal that
market impact costs cannot be calculated directly and suggested that these costs can be roughly
estimated by comparing the actual price at which a trade was executed to prices that were present
2

We note that effective October 14, 2016, upon the implementation of the new money market reform rules,
intermediaries will be asked to provide inter-day access to fund flow information for money market funds that are
subject to liquidity fees and gates to assist those funds in determining whether certain thresholds have been
surpassed that require a liquidity fee or gate to be imposed. If intennediaries are able to do so, we would like the
SEC to encourage intermediaries to leverage these capabilities so that all mutual funds could have access to intraday fund flow information.

17

fi.

T.Roweltice

INVES T WITH CONFIDENCE

in the market at or near the time of the trade. We feel that because these costs carmot be direct!y
calculated, and because they are not easy to quantify, the SEC should not require them to be
considered in the final rule.
We note that our SICAV funds, for example, are not required to take market impact costs
into consideration when determining their swing factor but are permitted do so, if they feel it is
applicable (for example, with certain illiquid securities). When setting their swing factor, our
SICAV funds primarily consider transaction costs, both explicit and implicit, such as broker
commissions or estimated dealer transaction fees paid by the fund, custody transaction charges,
tax implications, and foreign exchange costs. Unlike market impact costs, these costs are easily
quantifiable and do not require subjective prediction. In addition, for the same reasons as
applicable to the Proposal's six-bucket liquidity classification system, because market impact
costs are not easy to quantify, calculating them would require significant research and manual
manipulation, making swing pricing more labor-intensive and burdensome to perform.

Compliance Period
The SEC proposes that, as reliance on rule 22c-l(a)(3) would be optional, a compliance
period would not be necessary. However, some fund managers already have extensive
experience with swing pricing, while other fund managers will be approaching swing pricing for
the first time and, hence, be at a disadvantage. In order to provide a level playing field, we
believe the SEC should provide for a compliance date that is one year after the effective date.

******
We support the SEC's goal of strengthening liquidity risk management by open-end
funds and we appreciate the opportunity to submit our comments on this Proposal.
Thank you again for the opportunity to express our thoughts on this important topic.
Should you have any questions or wish to discuss our letter, please feel free to contact
Christopher Edge, Head of Equity Risk Management, at 410-345-2432, Fran Pollack-Matz,
Senior Legal Counsel, at 410-345-6601 , or Darrell Braman, Managing Counsel, at 410-3452013.

Sincerely,
ls/David Oestreicher
David Oestreicher
Chief Legal Counsel

18

I.

T.Roweftice

I NVEST WITH CONFIDENCE