Are Money Market Funds like CDOs?

A comparison of a money market fund to a collateralized debt obligation ("CDO") or structured investment vehicle ("SIV") may readily seem far-fetched, but given the complexity of risks and evolving regulations, one may appreciate the relevance of their similarities and resulting implications.

Why does such a comparison even matter, and why now? Recent amendments to Rule 17g-5, under the Securities Exchange Act of 1934, seek to improve the integrity of ratings issued by NRSROs (aka rating agencies) for structured finance products. What is a structured finance product under Rule 17g-5? According to the Federal Register (Vol. 74, No. 232):

The Commission intends this provision, which mirrors, in part, the text of Section 15E(i)(1)(B) of the Exchange Act (enacted as part of the Rating Agency Act), to cover the full range of structured finance products, including, but not limited to, securities collateralized by static and actively managed pools of loans or receivables (e.g., commercial and residential mortgages, corporate loans, auto loans, education loans, credit card receivables, and leases), collateralized debt obligations, collateralized loan obligations, collateralized mortgage obligations, structured investment vehicles, synthetic collateralized debt obligations that reference debt securities or indexes, and hybrid collateralized debt obligations.

In brief, a structured finance product includes static or actively managed pools of corporate loans and residential mortgages, among a number of other fixed-income securities and synthetic instruments.

The shortcomings of NRSRO-issued ratings of structured finance products is one of the commonly cited contributors to the Great Recession. However, NRSROs also play an important role in the credit quality of money market funds: Rule 2a-7 under the Investment Company Act of 1940 contains provisions which refer to NRSRO ratings for determining investment eligibility. Over the recent months, the SEC considered whether to allow fund managers to rely on ratings under Rule 2a-7. That debate led to the preservation of references to ratings in the investment process but added the requirement to use multiple NRSROs whenever possible and additional annual NRSRO quality assessments by the managers. According to a Mayer Brown Securitization Update (March 24, 2010):

Money market funds are regulated pursuant to rule 2a-7 under the Investment Company Act. Although recently amended to (among other things) remove some of its references to ratings, rule 2a-7 continues to rely to a substantial degree on ratings from NRSROs in defining minimum credit standards for fund investments. Some important rating standards in the rule are phrased in terms of specified ratings from the “Requisite NRSROs.” Ordinarily, “Requisite NRSROs” means any two NRSROs from a list of at least four that must be designated annually by the fund’s board of directors, but it can mean just one of the designated NRSROs if only one of them maintains a rating of the subject security.

NRSROs expanded their expertise beyond rating individual corporate bonds and structured finance products: they rated investment vehicles which held such rated instruments. Among such investment vehicles, the most often cited examples are CDOs and SIVs. Another less-publicized but more familiar example is the money market fund. How many money market funds do NRSROs rate? Using Standard & Poor's as an indicative NRSRO (throughout the remainder of this analysis), approximately 100 US money market funds were rated representing $283.9 billion of AUM, reported as of March 31, 2010. (The total AUM of US money market funds was approximately $2.8 trillion, according to the ICI as of June 30, 2010.)

The fact that certain types of investment vehicles are rated and invest in rated securities does not mean that they are all similar in credit risk posed to investors or sensitivity to the default risk in underlying investments. According to S&P in their article entitled "CDOs: An Introduction To CDOs And Standard & Poor's Global CDO Ratings" (June 8, 2007), CDOs cannot be deemed similar to mutual funds:

In a mutual fund, all investors share in the risks and rewards of the investments equally. In a CDO, the transaction is structured with different classes of notes, each having a different risk/reward profile. If any of the assets in the underlying asset pool default, the lowest note class (typically referred to as the CDO equity class) will suffer a loss. As losses increase in the asset pool, then the other classes of notes may also be affected.

This perspective differentiates CDOs from mutual funds on the key basis that mutual funds issue one class of security (generally an accurate assumption, with a caveat noted later) versus the multi-tranche liabilities of CDOs. However, this reasoning does not speak to credit and market risks from the underlying portfolio of fixed-income securities. In the same article, S&P further elaborates:

Investors can look at the asset portfolio to get a sense of the companies in the pool, how the companies are rated, and their line of business. Standard & Poor's also provides portfolio benchmarks of each CDO. The benchmarks include: Weighted average rating (WAR): The average rating on the companies in the pool; Weighted average maturity (WAM): The average life of assets in the portfolio; Default measure (DM): The expected annual average default rate of the portfolio; Variability measure (VM): The deviation around the average portfolio default rate; and Rated overcollateralization (ROC): The risk-adjusted collateral available to support a rated tranche.

In summary, S&P assesses the collective risks of underlying securities in a CDO based on their ratings, maturity profile, and expected default rates (with a certain amount of variability). Even if a CDO issues only one class or tranche of debt (much like a money market fund issuing one share class), the last benchmark, Rated Overcollateralization, is still relevant but simpler to analyze than for a multi-tranche structure.

The market-value CDO ("MV-CDO") is one type of CDO which serves as a useful and representative example of a rated investment vehicle. According to a criteria article entitled "CDO Spotlight: Criteria For Rating Market Value CDO Transactions
" (September 15, 2005), S&P provides the following definition of a MV-CDO:

Market value CDO transactions involve SPEs designed to purchase and actively manage a diversified pool of financial assets. Structurally, they are similar to cash flow CDOs in that their capital structures consist of a series of debt and equity classes. The primary difference between cash flow CDOs and market value CDOs is the nature of the risk passed from the pool of assets to the investor. As the name suggests, a market value CDOs risk is linked to the market value of the assets within it. The risk in cash flow CDOs is based on the pure credit risk of the assets, measured by the assets' ratings.

In addition to the risks attributed to the generic cash flow CDO, the MV-CDO includes a distinct exposure to fluctuations in the market value of underlying securities.

How does S&P characterize the risks of securities held by money market funds?

In a criteria article entitled "Fixed-Income Funds: Process And Overview" (February 2, 2007), S&P lists the following key risks within money market funds:

A Standard & Poor's Principal Stability fund rating, also known as a money-market fund rating, is a current opinion of a fund's capacity to maintain stable principal or net asset value. When assigning a Principal Stability rating to a fund, we evaluate the creditworthiness of a fund's investments and counterparties, the market price exposure of its investments, sufficiency of the fund's portfolio liquidity, and management's policies and overall ability to maintain the fund's stable net asset value (NAV) by limiting exposure to loss. In our view, funds that seek to maintain a stable NAV should be managed conservatively in regards to average maturity, credit quality, and liquidity and should follow well-defined guidelines and investment policies (such as those specified within SEC Rule 2a-7 guidelines).

In another criteria article entitled "Fixed-Income Funds: Market Price Exposure" (February 5, 2007), S&P further elaborates on key risks to the stable NAV of money market funds:

By far, the most complex part of money market fund analysis is judging a fund's sensitivity to changing market conditions. Absolute stability of net asset value (NAV) is a myth perpetuated by the amortized cost method of pricing securities. All fixed income securities are subject to price fluctuations based on the following: interest rate movements; maturity; liquidity; credit risk or perceived credit risk; and the supply and demand for each type of security. These factors are just as true for money market funds as for longer-term fixed-income mutual funds.

In summary, S&P assesses the collective risks of underlying securities in a money market fund based on their credit risk and credit quality (i.e. ratings), maturity profile, liquidity, and market price exposure. Do these risks sound familiar?

While there are several differences between CDOs and money market funds, once one compares how an NRSRO analyzes the risks of securities in the underlying asset pools, those differences begin to look more nominal and less substantive. Based on the above highlighted words, the risks underlying market-value CDOs and money market funds appear quite similar. Both investment vehicles predominantly, if not exclusively, hold fixed-income securities; hence, their risks should be assessed in a very similar manner.

The following diagram shows the similarities between the underlying securities in CDOs and money market funds.

This diagram classifies the structured investment vehicle (SIV) and CDO as comparable investment vehicles, both issuing multiple classes of debt securities. Several SIVs and CDOs issued short-term (commercial paper) tranches which, when rated A-1+ (or equivalent) by an NRSRO, were considered eligible for investment by money market funds. Interesting to note, those highly-rated short-term tranches of CDO/SIVs were held by some money market funds which were also highly-rated by the same NRSROs (too many examples exist to mention only a few judiciously).

As noted above, S&P stated that the single-class shares of money market funds distinguish themselves from multi-tranche debt securities typically issued by CDOs. Indeed, CDOs issued short-term and long-term debt tranches as well as an equity tranche (deemed the first-loss position). Do money market fund shareholders absorb any and all losses which may result from adverse performance in underlying securities, after deducting for expenses charged by the portfolio manager to act in its capacity? If you ask the shareholders of The Reserve Fund after Lehman Brothers filed for bankruptcy, the answer would be a resounding "yes". If you ask Federated, Charles Schwab, or BlackRock, based on their earnings reports, the answer would probably be "no" only because those firms (and many others) waived asset management fees in order to prevent their money market funds from "breaking the buck". Might such fee waivers be equivalent to a form of contingent capital injected into an otherwise loss-making business? (Posing this question does not seek to undermine the appropriateness of the fee waivers in the first place, rather to point out that capital was injected to avoid a loss to fund shareholders.)

In Comparison B, the equity tranches for CDOs and money market funds are added to Comparison A above.

At this point, one may still disagree with the premise that CDOs and money market funds, while regulated differently and usually sold to different types of investors, share similar types of risks from underlying fixed-income securities. When looking at the recent amendments to Rule 17g-5, this comparison can be extended to include a new dimension. Through Rule 17g-5, the SEC sought to mitigate the effect of conflicts of interest created by the manner in which NRSROs are engaged and compensated. According to the same Mayer Brown publication referenced earlier:

Among other changes, the Commission amended rule 17g-5 to facilitate unsolicited ratings from NSROs that were not hired by issuers, sponsors, or underwriters to rate particular asset-backed securities (ABS) and other structured finance products by enabling these non-hired NRSROs (Accessing NRSROs) to access the same rating-related information as “Hired NRSROs.”.

Do these conflicts of interest, deemed to have compromised the quality of past ratings on structured finance products, have any relevance to money market fund ratings? Building on Comparison B, the scope of Rule 17g-5 is highlighted in yellow in Comparison C.

Money market funds, already impacted by changes to Rule 2a-7 regarding the use of ratings, need to be mindful of potential, yet remote, side-effects of Rule 17g-5. With respect to the unsolicited rating of a structured finance product, Mayer Brown further elaborates:

When a fund relies on ratings from two of its designated NRSROs (which we would expect to occur in the vast majority of cases), no issues should arise from any unsolicited ratings. By definition, a security has the specified ratings from the Requisite NRSROs as long as any two of the designated NRSROs have provided the minimum ratings, regardless of what other ratings other designated NRSROs may have provided, whether on a solicited or unsolicited basis. However, if a fund seeks to satisfy the rating requirement based on just one rating, believing that only one NRSRO has rated a Structured Finance Product that the fund is buying, the fund could find itself unexpectedly holding an ineligible security because, unknown to the fund, another of its designated NRSROs rates the Structured Finance Product on an unsolicited basis. This would not generally require fund to dispose of the affected Structured Finance Product. Also, it should seldom occur, since most Structured Finance Products have two ratings from NRSROs, and the two NRSROs providing those ratings seem likely to be included in a purchasing fund’s designated group.

In summary, if an NRSRO issues an unsolicited rating for a structured finance product which is lower than the rating issued by designated NRSROs utilized by the money market fund portfolio manager, corrective action may be required. The changes to Rule 17g-5 are too recent to empirically assess whether this risk will be a material concern. If an NRSRO issues an unsolicited rating for a specific structured product security which is meaningfully lower than the ratings issued by other NRSROs, would the marketplace question or doubt the validity of the higher rating on the same security? Would the portfolio manager of a money market fund be able to argue against utilizing the lower unsolicited rating?

However, the real key take-away question is:

Should an NRSRO proactively issue an unsolicited rating of a money market fund, especially given any significant differences among NRSRO's in the perception of risks among the underlying securities?

The SEC has clearly targeted structured finance products as vulnerable to "ratings arbitrage". Since money market funds are not categorized as structured finance products (such as CDOs), how are their ratings (e.g. S&P's Principal Stability Fund Rating) protected from the conflicts of interest which prompted the SEC to amend Rule 17g-5?

Currently, a large share of US money market funds (possibly over 90%, based on S&P's coverage compared to ICI's statistics) are not even rated by S&P. Of the US money market funds rated by S&P, their average maturity of underlying securities (33 days as of March 31, 2010) is well under the SEC-mandated maximum of 60 days, indicative of the conservatism among those funds which request and pay to be rated. If a money market fund holds a portfolio which has an average maturity of 55 days, would the fund sponsor be less likely to pay an NRSRO for a rating (due to the higher risk of a potential downgrade should average maturity extend beyond 60 days)? In such a scenario, should an NRSRO issue an unsolicited rating?

After pondering this question, one might want (or not want) to consider an alternative question concerning accounting treatment: If money market funds are similar to CDOs, then should asset managers be required to consolidate money market funds onto their balance sheets under FAS 166/167 aka ASC 860? Under proposed rules, CDO managers already face a material possibility of consolidating their investment vehicles (see FASB Comment Letter Summary).

Additional Notes:

1. The above analysis is not exhaustive but rather seeks to raise awareness, through a concise discussion, of specific sources of potential (yet probably remote) risk in money market funds.

2. In order to comprehensively interpret the excerpts cited above, review the context from which the excerpts originated by clicking on the links to original source documents. Free registration may be required for certain source documents.

3. Certain words in the excerpts above are highlighted in yellow by Fundometry, not the original authors of the excerpts.

4. NRSROs other than S&P may have materially different perceptions of the sources of risk in structured finance products and money market funds.

Drilling Down into the Flash Crash, ETF Edition

Many media reports have commented on the severity with which stock prices collapsed during the afternoon of May 6, 2010 ("Flash Crash"). The most surprising casualty from the Flash Crash are ETF's. In general, ETF prices fluctuate consistently with the underlying index (for passive indexed, not active, portfolios) due to sufficient trading volume in the ETF itself and/or the ability for authorized dealers to use baskets of the underlying securities to profit from pricing discrepancies between the ETF and its constituent securities. ETF Trends wrote an article (referenced below) discussing the mechanisms for trading large blocks of ETF's either directly or indirectly via its constituent securities.

While ETF's justifiably raised concerns in light of the Flash Crash, the media has portrayed ETF's as a single asset class in this context. Some Wall Street Journal articles (referenced below) have presented examples of specific ETF's which suffered from extreme volatility that day. Although most of the trades which the exchanges cancelled on the evening of May 6 occurred in ETF's, many other ETF's did not incur enough price volatility to require trades to be cancelled. Few, if any, reporters have drilled down to determine which types of ETF's, and their sponsoring firms, performed better or worse on May 6.

This article attempts to reveal important distinctions among US-listed ETF's during the Flash Crash. In order to simplify the analysis, certain ETF's have been excluded. Those ETF's which utilize derivatives to achieve leverage (funds named with "ultra", "2x", "3x", and "double") are excluded in all comparisons because, by definition, their volatility is supposed to be higher than the general market, which would be confused with the exceptional volatility of un-leveraged ETF's on May 6. For similar reasons, inverse ETF's (funds named with "bear" and "short") are excluded. While the market was falling precipitously on May 6, inverse ETF's, by definition, should have not surprised investors in the same manner as long-only ETF's. Studying a single date of performance gives a limited scope of observations, but, nevertheless, readers can draw some very useful conclusions to help make future decisions on how to utilize ETF's in short-term trading or portfolio management strategies.

The simplest way to focus attention on the ETF's which exhibited unusually high volatility is to break down performance according to broad sectors. In the following table, we observe some sectors with severe price declines on May 6 regardless of high or low trading volume. The extreme magnitude of downside volatility is measured by how low prices fell in relation to the closing price that day ("Low-to-Close Difference", or "LCD"). For example, an ETF which traded as low as $1 but closed at $10 on May 6 would have an LCD of 90%. In order to better reflect the overall impact on market participants and investors, each ETF's LCD is weighted by the trading volume on May 6 to arrive at the Weighted Average Low-to-Close Difference ("WA-LCD"). Admittedly, a more accurate weighted average would consider the trading volume during the brief time during which most of the intraday volatility occurred, but that data requires more time to aggregate and process (perhaps a worthwhile objective for a future article on Fundometry).

As expected, ETF's invested in equities, including Asset Allocation, suffered from the most severe price drops. Large Cap (US) Equity ETF's performed better than Mid/Small Cap (US) Equity ETF's, which is consistent with the history of higher observed volatility in the prices of smaller companies. While the Asset Allocation sector exhibited the largest overall drop in price, that performance spans a broad range at the level of specific sectors (Moderate Allocation, Target Date, etc. as seen further below). Perhaps most significant, several sectors should be credited with avoiding the extreme price volatility of May 6: Taxable Bond, Municipal Bond, Global Bond, Currency, and Commodity ETF's. Finally, even though a broad sector may exhibit modest volatility overall, a single ETF within the sector may have performed much more poorly, as evidenced by the rightmost column "Worst Single ETF Low-to-Close Difference".

The next logical step in this analysis is to compare sectors which comprise the above broad sectors. The tiering of performance across specific asset classes and investment styles is quite apparent in the following table.

Clearly, the Small Cap Equity and Mid Cap Equity sectors rank among the worst performing sectors on the basis of the lowest trading price. Among the Small/Mid Cap group, the Small Cap Blend Equity sector held up by and far the best, with a weighted average lowest trading price 7.04% below the weighted average closing price. Industry-specific sectors suffered some excessive volatility, led by Telecom Sector Equity on relatively low volume. In order to judge the relative performance on most sectors, the Large Cap Blend Equity (which includes ETF's tracking the S&P 500) may be the benchmark. Sectors investing overseas performed relatively better. Among fixed income ETF's, the High Yield Bond sector exhibited the most downside volatility, trading 7.00% below the closing price on a weighted average basis. Relative to most sectors, the remaining bond sectors, as well as Commodity, Currency, and Target Date sectors, suffered limited downside volatility during the Flash Crash and exhibited the least negative WA-LCD figures.

Different ETF's sponsors offer products spanning various sectors. Therefore, when drilling further to the sponsoring firm level, the following tables each compare performance within a single broad sector. The term "ETF Complex" includes the ETF's offered by one firm, or within one family, which fall within the specified broad sector, and each WA-LCD is computed accordingly. The first table below compares ETF Complexes which offer Large Cap Equity portfolios.

At the ETF Complex level, trading volume shows a stronger correlation to the WA-LCD metric. ETF's from FaithShares and Grail have the lowest trading volume among their peers, perhaps to the extent that these ETF's did not trade much during the most volatie period on May 6. The leading ETF Complexes by volume - PowerShares and State Street SPDR's - performed consistently with the broader market primarily because of their high-volume products - QQQQ and SPY, respectively. (Remember, WA-LCD is weighted by trading volume on May 6.) The next most-traded ETF Complex - iShares - suffered a disappointing -44.82% weighted average drop, relative to the closing price on May 6, due to underperformance in their S&P 500 and Russell 1000 offerings. The remaining ETF Complexes which suffered disproportionate price drops also displayed very modest trading volumes relative to peers.

The comparison of ETF Complexes proceeds with the Mid/Small Cap broad sector.

Quite contrary to its performance in the Large Cap Equity broad sector, iShares exhibited the smallest overall price drop during the Flash Crash due to the high trading volume of IWM, which tracks the entire Russell 2000 index. The second-most heavily traded ETF Complex, State Street SPDR's, also held up relatively well against its peers primarily due to the high trading volume of MDY, which tracks the entire S&P MidCap 400 index. Neither of the next two most actively traded complexes - Vanguard and PowerShares - offered a specific Mid/Small Cap ETF with high volume and strong performance, resulting in significantly negative WA-LCD levels.

According the to following table, State Street SPDR's well-known family of sector ETF's, based on the S&P 500 constituents, performed respectably compared to their peers, probably due to their large share of trading volume among sector-focused equity ETF's. Van Eck's Market Vectors ETF's also suffered limited volatility, due to the outperformance of some commodity-sector equity ETF's within that complex. Likewise, the ETF Complex with the least negative WA-LCD among sector-focused products was Jefferies, with its equity portfolios based on Thomson-Reuters/Jefferies CRB indices.

On that note, the next comparison drills down into the Commodity broad sector, consisting of ETF's and ETN's which track physical commodities. Unsurprisingly, this group of ETF Complexes suffered modest declines during the Flash Crash. The PowerShares Deutsche and UBS E-TRACS complexes suffered the largest price declines, mostly due to isolated drops in specific ETF/ETN's over low volume. The most traded complexes within the Commodity broad sector, iShares and United States, exhibited consistent performance across their specific ETF's. For ETF/ETN's which give investors exposure to physical commodities, the volatility on May 6 can hardly be described as a Flash Crash.

In addition, the performance of ETF Complexes which track taxable bond portfolios exceeded the performance of their equity peers, as seen in the table below. The overall performance of PowerShares, State Street SPDR, and iShares complexes suffered from exposure to high-yield bond portfolios, specifically PHB, JNK, and HYG respectively. Otherwise, price declines in taxable bond ETF's throughout all of these complexes were modest. As should be expected, underlying ETF portfolios with longer average durations mostly exhibited greater price declines. Very few signs of a Flash Crash existed for ETF's which track taxable bond indices, and even more so for municipal bond indices.

(Similar tables for other broad sectors are available by popular demand. Interested readers may submit requests in the comments section below.)

The above comparisons clearly show that several ETF's, and some entire sectors, did not suffer from excessive volatility on May 6. Undoubtably, volatility across all market sectors was higher than normal that day. Even within complexes, the worst WA-LCD levels show some outlier downside activity in otherwise sheltered sectors. However, the market should avoid generalizations when assessing liquidity and price discovery of ETF's. Depending on its underlying portfolio, an ETF possesses a unique set of risks which may help or hurt performance during a Flash Crash scenario.

During the Flash Crash, why did some ETF's suffer less downside price risk than others, even among those which track similar indices or hold similar portfolios? The above discussion does not provide many insights which would help answer that question. Based on a yet-to-be-published analysis, no single set of factors would have predicted which ETF's outperformed on May 6. Yet to be released data covering trading activity in May might shed further light. Stay tuned.

Related articles:

How to Trade Large Blocks of ETFs Efficiently (ETF Trends)

Danger: Falling ETFs (Wall Street Journal)

A 'Flash Crash' Lesson (Wall Street Journal)