Morningstar’s Defined Contribution Outcomes Model: How to Better Evaluate Retirement Plan Design

When you’ve worked at a big company like Morningstar for as long as I have, you become a little bit skeptical when coworkers start tossing around the word “innovative” to describe a new product or service they’ve dreamed up. Nine out of 10 times, the “innovation” is cool, but far from being a game changer. So, when the research team within the Morningstar Center for Retirement & Policy Studies told me about this “innovation” called the Defined Contribution Outcomes Model, or DCOM, I set my expectations accordingly.

My first question was, "What is DCOM? Sounds like an archaic Microsoft operating system that nobody knows how to code to anymore.”

My second question was, "What makes this so innovative?"

DCOM is a highly sophisticated simulation model that our researchers can use to predict retirement outcomes for individuals or groups of individuals.

The model took our Morningstar Retirement research team about four years to build. It's powered by data from our recordkeeping clients, including more than one million participants and thousands of different plans that represent about 42 different plan types. It essentially lets us test how changes—large or small—to plan design, savings rates, and investment outcomes can affect retirement outcomes for different participant populations.

Let’s say you want to test whether participants would benefit from switching from a target-date fund to managed accounts. DCOM evaluates who could and couldn’t benefit, and under what plan design conditions. This type of analysis may then be used by a plan to see if switching the default makes sense, and to isolate which cohorts within their plan could benefit the most from the switch. We can also determine if the effect on outcomes is the same whether the plan has, for example, auto-escalation or not. Additionally, we’re able to determine the impact based on the plan’s matching formula.

Unlike other models that simulate retirement outcomes, DCOM accounts for participant behavior, market volatility, and financial lifestyle. As anyone who has been in the workforce for any significant period can attest, the journey from starting out in that first job after college to the day you retire is far from a straight line. Like investment markets, there's a degree of volatility that occurs in a person’s career—from layoffs and salary reductions to challenges paying for a kid’s college education to paying down credit cards and mortgages. As a result, employment is never consistent, and contribution rates can fluctuate, sometimes widely and unpredictably.

Many traditional models rely on static assumptions that savings rates and employment stay constant. This typically means retirement outcomes from this research are mostly best-guess estimates. DCOM is perhaps one of a few models in existence that’s based on real participants and their data accounts for market and lifestyle volatility, making our results more granular and arguably more accurate.

Here's a good analogy for how DCOM works in comparison to more traditional models and methods. Think about those more traditional models as being a Rand McNally map crumpled up in your glove compartment. You could use it to determine the approximate mileage from point A to point B. Then, based on how fast you planned to drive and whether you needed to navigate any major metropolitan areas, you might come up with the amount of time it would take—give or take a few hours. Now think of DCOM as Google Maps. The model will factor in GPS data from users with historical traffic patterns, official speed limits, and road characteristics. It can also consider current traffic density, accidents, and road closures. Plus, you can use it to test different routes and the impact on your journey. The end result is a much more accurate ETA, almost down to the minute.

Our Retirement Center’s mission is to determine ways that the industry and policymakers can improve the retirement system in the US. 

So, for example, let's say that the federal government mandates that all DC plans in this country implement auto-escalation features. We can test how that would impact different groups across plans and plan types. We can get very specific in our findings, looking at outcomes by age, gender, salary, education level, etc.

DCOM can help plan sponsors determine whether the ROI is worth it when modifying their plan designs, like a match formula. The model may give that plan an idea of how this change could affect their participants when it comes to retirement outcomes. Again, we can perform a more detailed analysis of the impact, looking at specific populations within the plan.

Additionally, an advisor, consultant, or recordkeeper may use it to stress-test a plan they’re working on. For instance, they could see how the current investment lineup would fare across various market scenarios, as well as how changes to the lineup and defaults would impact the participants in the plan. This can enable them to make more evidence-based decisions when offering advice or building a plan.

The development of DCOM is a significant advancement in our ability (and the industry’s ability) to run more accurate and true-to-life simulations to test plan and investment changes.

It moves the industry from simplistic, average, and static assumptions to a framework that's dynamic and multidimensional. This will result in more informed decisions when it comes to designing retirement plans—from the investments offered to the underlying plan features. It’s also an accurate way to test policy and regulatory changes, with the hope that they’ll lead to better outcomes.

We currently have no plans to externalize our DCOM. Along with using it to power our own research, we’re running custom reports for recordkeepers who have agreed to provide us with access to anonymized plan and participant data on their platforms.

If you want to get a sense of what we can do with DCOM, I recommend taking a look at the latest report on the value of managed accounts from Morningstar’s Jack VanDerhei and Spencer Look—the masterminds behind the DCOM model. This paper is the first of many that will use DCOM to test various assumptions about the value of managed accounts.