The Overhead Myth Debunked: How AI Reveals the Truth About Nonprofit Overhead

Attention nonprofit and philanthropic professionals: It's time to rethink the "overhead problem." Our sector has been trapped in a vicious cycle of unrealistic expectations and chronic underfunding for far too long. Despite years of research, there has been no real change in how donors are willing to fund overhead. The root cause of this issue is the chronic underinvestment in overhead expenses, which leads to the nonprofit starvation cycle.

In their influential 2009 article "The Nonprofit Starvation Cycle," Ann Goggins Gregory and Don Howard argue that funders' unrealistic expectations for low overhead costs lead nonprofits to underreport and underinvest in essential infrastructure. This, in turn, reinforces funders' belief that nonprofits can operate with minimal overhead, perpetuating a vicious cycle of underfunding.

The consequences of this cycle are dire. As Lecy and Searing's 2015 study "Anatomy of the Nonprofit Starvation Cycle" demonstrates, the pressure to minimize overhead leads to deep cuts in administrative expenses, hindering nonprofits' ability to invest in the systems, training, and staff needed to serve their missions effectively. Over time, this chronic underinvestment erodes organizational capacity and undermines long-term sustainability.

But now, with the power of AI and machine learning, we can study the complexities of this difficult question and determine the right level of overhead more fairly. Enter the Equitable Impact Platform (EquIP), developed by BCT Partners (BCT). EquIP combines IRS 990 tax data from all U.S. nonprofits with the Census Bureau's American Community Survey data, applying cutting-edge AI algorithms to match nonprofits based on their characteristics, size, community well-being, and access to funding. This precision causal modeling approach, powered by AI, provides a more equitable way to assess and determine the right amount of overhead for each organization.

Precision causal modeling is BCT’s sophisticated AI-driven method that goes beyond simple correlations to identify the causes behind an organization's success. It uses machine learning algorithms to find "matched groups" of similar nonprofits - ones that operate in comparable contexts, serve similar populations, provide similar programming at a similar scale, and have access to the same types of resources. By studying the natural experiments whereby comparable organizations choose different overhead levels over a four-year period, AI can test these counterfactual differences and isolate the effect of different overhead levels on four-year program output, controlling for all the other factors that might influence success.

This approach is revolutionary because it allows us to see the real impact of overhead spending, free from the confounding variables that have clouded this issue for so long. With precision causal modeling, we can finally answer the question: for nonprofits like this one, operating in this specific context, what level of overhead is truly associated with program sustainability and growth? The result is a much more nuanced, accurate, and equitable understanding of how overhead affects nonprofit performance, tailored to the unique realities of each organization.

So, what's the key takeaway here? It's simple: the right level of overhead is not a one-size-fits-all number. It depends entirely on the unique context of each nonprofit. As Todd Rose compellingly argues in his book "The End of Average: How We Succeed in a World That Values Sameness," we must move beyond blanket solutions and embrace a more nuanced approach.

EquIP's cutting-edge, AI-driven analysis reveals that a nonprofit's mission and operating model are critical factors in determining the appropriate overhead level. Let's take youth sports leagues as an example. These organizations often rely heavily on volunteers to deliver their programs, which means that their overhead costs - things like fundraising, technology, communication, equipment, etc. - make up a larger percentage of their total expenses. In fact, EquIP's data shows that these leagues are more likely to thrive with an overhead ratio as high as 60%, 70%, and yes, even as high as 80%.

Now, some funders might balk at that number. But when you think about it, it makes perfect sense. Volunteer-driven organizations don't have a lot of program costs because their volunteers are generously donating their time. So, a higher overhead ratio isn't a red flag; it's a reflection of their operating reality. In fact, if donors insist that all their contributions go directly to programs, they could actually starve these leagues of the resources they need to function effectively. EquIP's analysis shows that when youth sports organizations spend a large majority of their funds on overhead, they are much more likely to sustain and grow their programming over a four-year period. That's right - investing in "overhead" is what allows them to serve more kids, more effectively, for years to come.

The MacArthur Foundation's recent research, where BCT Partners used a similar AI-based approach, underscores these findings. By analyzing Form 990 data with machine learning, they found that the minimum indirect rate associated with financial health varied across different types of nonprofits. This led the foundation to revise its indirect cost policy, setting a 29% minimum rate for all grantees to help them invest in essential infrastructure.

EquIP takes this a step further, using AI-powered causal modeling to predict the right level of overhead for each nonprofit to achieve sustainability and program growth over a four-year period. The results confirm that "it depends" - there is no magic number that applies to all organizations.

So, what does this mean for funders and Grantmakers? It's time to embrace a more nuanced, data-driven approach to overhead, made possible by AI. Rather than imposing arbitrary limits, use big data, causal modeling, and AI-powered tools like EquIP to understand each nonprofit's unique needs based on their work, community, and context. By doing so, we can break the nonprofit starvation cycle and invest in the infrastructure that enables nonprofits to thrive, not just survive.

The overhead myth has persisted for far too long. But with the power of AI, big data, and precision causal modeling, we now have the evidence to debunk it. Let's use this knowledge to create a more equitable and effective nonprofit sector, one that has the resources to truly achieve its mission. The future of our communities depends on it.

LEARN MORE:

• Check out this BCT Partners video about its Equitable Impact Platform (EquIP).
• Listen to Episode 9 of The Nonprofit Fix Podcast to learn more about the nonprofit overhead problem and the AI solutions.
• Everyone should read Todd Rose’s book, The End of Average, to learn more about the suffering caused by using ‘averages’ to make decisions.
• Contact us, BCT Partners, to learn more about our data analytics work, EquIP, and Precision Causal Modeling
• Click on the embedded links in this article.