Analyzing Separately Managed Accounts

Journal of Financial Planning: November 2011

Mark W. Riepe, CFA, is a senior vice president at Charles Schwab & Co. Inc. and president of Charles Schwab Investment Advisory Inc. in San Francisco, California.

Like moths drawn to a flame, researchers tend to direct their efforts toward subjects that have the richest sets of data. This explains why the majority of serious studies on investment managers have focused on mutual funds.

This focus on mutual funds is at odds with how investors choose to use the various forms of delegated money management. For example, the aggregate amount of dollars invested in mutual funds is similar to that invested in separately managed accounts (SMAs).¹

With the advent of high-quality data sets² that document the returns and characteristics of separately managed accounts, the disparity between mutual fund studies and SMAs is being addressed. My colleagues Jim Peterson, Michael Iachini, and Wynce Lam (hereafter referred to as “PIL”) provide an excellent example of the new work in this area.³ The question they seek to answer in their work is simple: are there characteristics of SMAs observable today associated with future performance?

Before we get to the results of their work, there are three major differences between SMAs and mutual funds.

1. Ownership of the securities held by the manager. In a mutual fund structure, the securities are owned by the fund and investors own shares of the fund. In an SMA the account holder owns the securities.
2. Expenses. The costs of operating a mutual fund are deducted from fund income so that all shareholders in a share class pay the same fee (expressed as a percentage of assets invested). For SMAs, asset management fees, other administration fees, and commissions can vary for each client.
3. Account restrictions. With SMAs, the account holder can request that certain restrictions be placed on the account (for example, not to hold a certain stock). Mutual fund shareholders cannot do this. Because of the ability to somewhat customize the holdings of the account, SMA managers report “composite” returns. These are an average of the returns experienced by clients who invest in a particular strategy.

Results of SMA Research

PIL studied 3,081 domestic equity strategies from 1991–2009. The following variables were assessed for their correlation with future performance (defined as the alpha of the strategy’s composite over the next 12 months), and here’s what they found.

• Past Alpha. Alpha was derived from a three-factor model in which the factors were market performance, growth/value, and market capitalization. An alpha estimate from a four-factor model (adding momentum) was also used. To help distinguish whether different past periods of performance had different levels of influence over future returns, short-term, medium-term, and long-term alphas were calculated. PIL found that past alphas were positively related to the alpha generated over the next 12 months, irrespective of whether alphas were estimated with or without the momentum factor. In addition, the short-term alpha was more associated with future performance than the medium-term alpha. The long-term alpha was meaningful for large-cap strategies in the database, but not for small-cap strategies.
• Active Share. This is a measure of how much the manager’s historical portfolio returns deviate from those predicted by the alpha model discussed above. The idea behind this variable is that managers whose returns are close to the returns expected after taking into account market risk, style risk, and capitalization risk are less likely to generate future alpha. PIL found funds that had higher active share tended to have higher future alphas.
• Portfolio Turnover. Portfolio turnover refers to the rate of trading activity in the portfolio. This is associated with higher future alpha when alpha is estimated with the momentum factor. However, include the momentum factor and the significance of turnover drops out. These results suggest that the positive relationship between lagged values of turnover and alpha is primarily driven by managers implementing a momentum strategy.
• Assets Under Management. This variable refers to the assets in the strategy. Some analysts believe more assets in a strategy hinder the manager’s ability to add alpha. If this is the case then a negative relationship should exist between assets under management and the alpha generated by the strategy. This hypothesis is supported even when the strategies are divided into large- and small-cap samples.
• Number of Accounts. This refers to the number of accounts in the composite of the strategy. The hypothesis that motivated the inclusion of this variable is that as the number of accounts grows, an asset management organization may have to hire more portfolio managers, traders, and client-servicing staff, and this growth may cause administrative stress. However, in PIL’s empirical work they found that the relationship between the number of accounts and future alpha appears weak.
• Cash Inflow. When investors pour new money into a strategy it can interfere with the management of the portfolio. Past work at Schwab found that modest amounts of cash inflows can be accommodated in most situations, and only extreme levels of cash inflows are likely to have performance implications. To capture this effect, PIL focused on large cash inflows. They find confirmation of this hypothesis for small-cap strategies, but not for large-cap strategies.

PIL concluded that when evaluating SMAs, past performance does matter assuming the performance is properly adjusted for various sources of risk. However, there is more to the story. Those managers willing to be more active had better returns, as did managers who didn’t get too large in terms of assets invested in their strategy. Finally, when it comes to small-cap strategies, pay special attention to those that have recently experienced large cash inflows.

PIL didn’t include expenses even though one of the most persistent results in this type of literature is that expenses are associated with lower returns. Expenses weren’t included in this work because expenses vary from client to client. However, given the overwhelming evidence from the mutual fund world that expenses matter, it seems logical to conclude that expenses would show up as a meaningful variable in this work if the data were available.

A final note is we shouldn’t necessarily assume the sorts of relationships identified by PIL remain constant over time. In fact, an advantage of the long period studied in this work is that the period can be divided into smaller periods to see whether impact of variables on future alpha remains stable over time. PIL did such an analysis and generally found that the results in the second half of the sample period (2001–2009) were directionally the same as the results in the first half (1991–2000)—though the strength of the results was weaker.

Endnotes

1. Cerulli Associates. 2009. “Cerulli Quantitative Update: Managed Accounts 2009” (October).
2. The data used in this study are from Informa Investment Solutions’ PSN Investment Manager Database. The PSN database is a comprehensive, global database consisting of approximately 2,000 organizations representing more than 10,000 investment strategies.
3. Peterson, James D., Michael Iachini, and Wynce Lam. 2011. “Identifying Characteristics to Predict Separately Managed Account Performance.” Financial Analysts Journal (July/August): 30–40.