Sheryl Rowling's Practice Wise: The Smart Way to Use Model Portfolios

Sheryl Rowling's new Practice Wise column is available first in Morningstar Office Cloud.

Advisors tend to resist whenever they hear a recommendation for standardization. Yet, standardizing some aspects of their practices can improve their client experience. One such area is the use of model portfolios.

I’m a big believer in Modern Portfolio Theory. Most advisors are familiar with the 1986 study by Gary P. Brinson, Randolph Hood, and Gilbert L. Beebower that concluded over 90% of a portfolio’s expected performance can be explained by asset allocation. It certainly seems like that’s where we should spend our time.

I also believe that we are capable of creating our own models. Rather than cookbook off something on Google or opt for pricey outsourced solutions, advisors have other options such as utilizing models from Morningstar’s Model Marketplace or simply and effectively creating models that are tailored for their clients with Morningstar tools.

My firm has been creating its own models for decades. Here’s a roadmap behind the process for creating and using model portfolios.

Before building any type of portfolio strategy, advisors need to determine how they want to manage investments. Will they actively manage portfolios or go with asset allocation? Will they use standardized models? Will investments be in individual securities, mutual funds, or exchange-traded funds?

Then when it comes to constructing the models, there are several points to consider:

• How many models do you want?
• How many asset-type "slices" do you want in each model?
• How will you handle exceptions?
• Will you offer different sets of models—that is, environmental, social, and governance or small account portfolios?

I’ve found that the best approach is to avoid overcomplicating things, starting by limiting how many models you offer.

It may be tempting to think that more models are needed to address the myriad client needs. But there’s not a significant risk/return difference between a 75/25 portfolio and an 80/20 portfolio. So, why create multiple models with small variations? Also, limit how many asset classes you include in each model. A 2% allocation is not going to significantly impact the portfolio, but it will result in more rebalancing trades, and this eats into returns.

At my firm, we use only five model portfolios. We don’t believe in 100% bond portfolios, so we don’t offer them. We believe each client can be served well by one of these stock/bond allocation models.

We also limit the number of “slices” held within a model. It’s a delicate balance: You want to have enough slices to achieve a fully diversified portfolio, but not so many that management becomes overly cumbersome. In addition to transaction costs, working with many small allocation pieces can be a nightmare to oversee. A simpler model can provide full diversification without the complexity.

Part of this question involves deciding whether adding in a particular asset class is worthwhile. Probably the clearest example is emerging markets. If emerging-markets equity will only represent 3% of the portfolio, will the diversification benefit outweigh the stress clients feel when this asset class experiences a 40% decline?

My rules for asset-class inclusion are as follows:

• Include U.S. and foreign
• Include equities and bonds
• Include large and small
• Limit alternatives
• Avoid overly volatile
• Avoid tiny allocations

Whenever using any type of optimizer, unconstrained results will sometimes create unusable models. In fact, I’ve seen “efficient” portfolios made up of just two asset types like real estate and international bonds.

To some degree, creating models involves both science and art. In my practice, we place minimum and maximum limits to create valid, more palatable models. For a 60/40 portfolio, we typically limit real estate to no more than 10% and not less than 4%. We limit U.S. core bond strategies to no more than 30% and not less than 15%. Is there a scientific reason for these limits? Not really. But we need to consider what will make sense in the long run and what will feel right to our clients--and that’s the art.

When trying out different iterations in the optimizer, the questions of “Should I include or exclude this asset class?” or “Should I increase or decrease the allocation to this asset class?” depend on the following analysis:

• Will the change increase return without increasing risk?
• Will the change decrease risk without decreasing return?
• Will the change increase return and decrease risk?

If any of these questions can be answered yes, then the model change is justified.

From a practical standpoint, some clients would seem to warrant an exception. I believe that advisors should not allow exceptions. In fact, allowing exceptions could be detrimental to the client.

Essentially, there are two types of exceptions--a client wants to exclude a slice that’s in your model or a client wants to invest in something that’s not in your model. If you, as an advisor, go along with this, you’re complicating your life and now setting yourself up for returns that don’t equate to a model benchmark.

If a client owns many rental properties and then doesn’t want real estate in their managed portfolio, you might suggest a portfolio with a lower equity allocation. If your portfolio calls for international bonds and the client doesn’t want to invest in them, it’s up to you to educate the client on how the bonds work within the asset allocation. If a client wants to invest in gold and you don’t include gold in your models, you can explain why or simply have the client invest in gold outside of your managed portfolio.

Finally, although I personally believe in the five-model approach, I also believe in options. The first option is ESG. Many clients want their investments to be in sync with their beliefs. Rather than work with exclusionary screens, which can differ for each client (some clients want to avoid companies that do animal testing, others want to avoid defense contractors), we prefer funds with positive ESG ratings. Although this does not affect the five-model approach, it does affect the choice of funds. (A prediction: In the long run, we might utilize ESG factors in all offered portfolios.)

The other option is for small accounts. In the old days, we would simply use asset-allocation funds for small portfolios. Although this was efficient, when a small account would “graduate” to a full allocation model, we had a dilemma: Do we keep the allocation fund even though it’s not part of the full model or do we sell and incur tax for the client? Our solution to this issue was to create “building block” models. For example, the smallest portfolios would contain only two basic asset classes: U.S. large cap and U.S. core bond. As the portfolios grow, we add in new slices until we get to the ultimate full allocation.

To do the best job for all your clients, you must be able to create a degree of standardization within your practice. Even with a limited number of models, clients will have individualized experiences--because they have different amounts held in taxable, Roth, and retirement accounts; they have different tax situations; they have different goals; and they are in different life stages. By standardizing your models, you will have more time to focus on each client’s personal financial planning needs.