CUSIPs vs. Characteristics: Managing Fixed Income Portfolios to a Model

Portfolio managers use model portfolios to structure and update positions across their clients’ portfolios, for both equities and fixed income. The benefits of using a model are clear: it communicates a firm’s or home office’s best thinking on the most advantageous exposures and how to achieve them. That allows advisors to serve dozens, hundreds, even thousands of clients without having to individually determine which exposures are desirable for each one, one at a time.

Furthermore, when new cash comes into an account there is no ambiguity about how to invest it—it is allocated to keep the portfolio in line with the model. As markets shift and the model portfolio is updated, PMs can quickly determine what needs to be done to align each affected account with the latest thinking as revealed by the revised model.

However, given that equity and fixed income markets are vastly different, implementing the model portfolio approach the same way for equity and fixed income portfolios is frankly illogical and impractical. In fact, rigidly adhering to a process based on how equity markets work can detract from results for fixed income. However, in the past, PMs have been limited to the ways they can manage fixed income models. Here, we explain why, and describe a superior approach to applying model portfolios to fixed income.

Drawbacks to a CUSIP-driven approach for fixed income

When managing to a model portfolio of equities, it is a reasonably simple exercise to match each client’s equity holdings to the model portfolio’s positions, CUSIP by CUSIP. Given the structure of the equity market, we can see why this makes sense; however, that approach doesn’t fit the nature of bonds or the way fixed income markets work. With bonds, the goal should be to match each client’s portfolio to the model’s characteristics, optimally achieving the desired exposures without having to hold the specific CUSIPs in the model portfolio.

While some advisors do try to manage their clients’ fixed income portfolios using the CUSIP-driven paradigm developed for equity markets, this approach is (forgive the cliché) like trying to fit a square peg into a round hole. It doesn’t work well, and if you try to force it, something is likely to break.

Here’s why CUSIP-based models are not optimal:

• Sourcing securities: With equities, it is easy (with a few exceptions, such as micro-caps or frontier markets) for portfolio managers or traders to source all of the stock CUSIPs in a model portfolio, in the correct proportions, across a large number of accounts. With fixed income, this is far from easy. For starters, as there is no centralized exchange for trading corporate bonds, sourcing the specific CUSIPs in order to match every client’s holdings to the model portfolio’s, in the amounts needed, is challenging and time-consuming at best. That is the reality, whether a trader or PM uses centralized platforms such as MarketAxess or Tradeweb, or works with individual dealers, or a combination of sources. That means an advisor’s relationships with various broker-dealers can have an outsized influence on how quickly you can align your clients’ portfolio with the model portfolio, whenever new bonds are added to the model. Additionally, exact CUSIPs can be difficult to source, whether it’s a specific maturity or munis in which the secondary market is not as robust.
• Liquidity: The flip side of trying to buy the specific bond CUSIPs held in a model portfolio, a PM using the CUSIP-based approach must also sell the bonds as they are dropped from the model, across all relevant client portfolios. That could take days or even weeks, putting some clients at a disadvantage to others if interest rates and spreads move while those trades are being arranged.
• Built-in price insensitivity: In both buying and selling bonds, being limited to specific CUSIPs means the PM or trader must be a price taker, willing to buy or sell at whatever price is available. And there is a high likelihood that many of the transactions needed to bring portfolios in line with the model when the model changes, or when one or more portfolios has excess cash, would involve trading odd lots. All of this means buying and selling at whatever price a counterparty offers, affecting yield and total return.
• Restricted buy lists: As bond issues are limited in size, it can be difficult to find certain CUSIPs, especially if a big slice of an issue is held in buy-and-hold portfolios. Recognizing this, model portfolios usually restrict their holdings to large, recent issues. That limits the opportunity set for portfolio managers, removing an important way to generate value by choosing bonds with the desired characteristics at a better yield, perhaps exploiting a size premium for small issues. The CUSIP-based approach to tracking a model portfolio would prohibit this coloring outside of the lines.
• No personalization or tax optimization: Replicating a model portfolio at the CUSIP level does not allow advisors to personalize clients’ portfolios. There is no mechanism for excluding certain industries (such as tobacco or casinos) or specific names that a client may see as unfavorable for whatever reason. Furthermore, bonds are included in a model portfolio because they were appealing at the time they were added. But when an investment thesis has played out and the bond is no longer priced attractively, it exits the model portfolio. That can force PMs to sell bonds from clients’ portfolios at an inopportune time, with tax consequences (either gains or losses) that affect the client but not the model portfolio.
  
  

Matching a model bond portfolio using characteristics

A better alternative to tracking a model fixed income portfolio at the CUSIP level is to create portfolios that match the idiosyncratic characteristics of the model without being slavishly tied to the specific CUSIPs it holds. This approach overcomes all of the problems described above.

Some of the obvious, high-level portfolio characteristics to match using this approach include duration, average quality sector weightings, and yield-to-worst (or yield-to-maturity, if you prefer), but that is just painting with broad strokes.

Here are some characteristics to monitor to ensure clients’ portfolios track the model closely:

• Term structure risk: To match the of the model portfolio, PMs could align the key rate durations of clients’ portfolios with the key rate durations of the model portfolio, or match yield curve exposures using duration buckets (e.g., the percentage of holdings in the 0-1 year, 1-3 year, 3-5 year duration buckets, and so on).
• Spread risk: To go beyond matching the percentage held in a given sector to a model portfolio, PMs can use the concept of Contribution to Duration. This combines the weightings and durations of the bonds in each sector to capture the exposure to changes in spreads on a portfolio (long duration bonds have much more spread risk than short duration bonds)
• Quality: Rather than simply matching the overall quality of a client’s portfolio to the model, aligning the weightings for each credit rating, from AAA to BBB- (or lower) would ensure better tracking.
• Issuer risk: Restricting portfolios to a list of approved issuers based on the model’s holdings without insisting on owning specific CUSIPs provides much more latitude and makes sourcing bonds much easier. It simply makes sense that if a bond issued by XYZ Inc. is part of a model portfolio, PMs should be indifferent as to which CUSIP for that issuer the portfolio holds, assuming all of the other aggregate characteristics mentioned above are in line with the model. The CUSIP-based approach doesn’t offer this flexibility.
  
  

Technology enables PMs to reach the goal

The idea behind the characteristics-based approach is to match a model portfolio without being forced to accomplish its objectives in exactly same way across all client portfolios.

For example, to extend duration a model portfolio builder chooses specific bonds to buy and sell. Advisors who use CUSIP-matching must make the same changes across their portfolios, regardless of tax considerations, liquidity, etc. With a characteristics-based approach, there are many ways to extend duration while maintaining the same yield curve exposures, spread risk, and credit quality. And, PMs can manage tax consequences and incorporate client preferences, and traders can avoid trading odd lots that would be needed to buy the required number of specific CUSIPs.

A feature-rich optimizer, such as the one IMTC offers, can show how to shorten or extend duration, achieve target yield curve exposures using key rate durations and/or buckets, manage sector weightings using contribution to duration, and match quality ratings based on a model portfolio. This approach also allows PMs to minimize gains or losses and incorporate clients’ preferences (for example, by excluding certain issuers) across large groups of portfolios, using a realistic buy list of available issues.

Using an optimizer to match a model portfolio’s detailed characteristics achieves the goal of following the model portfolio builder’s best ideas without going off script. PMs achieve the desired characteristics with none of the limitations and disadvantages of the CUSIP-based approach.

Another key aspect of technology that is important in managing a model portfolio is the ability to set compliance rules. This automatically keeps portfolios in line with the model when using the optimizer and can alert PMs when a portfolio is out of line. In particular, IMTC’s functionality enables fixed income managers to set a portfolio in line with a model and minimizes the logistics burden of managing these rules.

Explore IMTC’s portfolio management functionality >>

A clear verdict: match characteristics, not CUSIPs, for fixed income

Replicating a model portfolio CUSIP-by-CUSIP for a client’s fixed income exposures may seem easier than using a characteristics-based approach because specifying individual CUSIPs to buy or sell requires little thought and may piggy-back onto the process in place to manage equity portfolios against a model. But what seems like an easy approach comes at a great cost in terms of flexibility and performance.

Given the growth in popularity of separately managed accounts, using model portfolios has become even more prevalent. Assuming a firm has the right tools in place, using the right metrics and a capable optimizer to match client portfolios to a model and make adjustments over time is far superior to a CUSIP-matching approach.