Researchers Explore State Expenditure Forecasting Challenges

Editor's note: This Q&A was updated on September 24, 2024 to correct the interviewees' names and titles.

Long-term revenue and spending projections can help states determine whether their budgets are on a sustainable path—or whether, instead, they can expect to face structural budget gaps. But although 30 states project revenue and spending at least three years into the future, the rigor, accuracy, and policy relevance of these projections vary.

To better understand the challenges state analysts face in producing long-term expenditure forecasts, and to explore potential solutions to these challenges, The Pew Charitable Trusts commissioned a report from the Government Finance Research Center (GFRC) at the University of Illinois Chicago.

As part of its research, the GFRC team spoke with staff from the offices that publish long-term expenditure forecasts in Colorado, Louisiana, Minnesota, New York, North Carolina, South Dakota, and Washington. (The team also collected insights from Tennessee, which does not publish long-term expenditure forecasts but produces them for internal use.)

The resulting report, “Planning for Tomorrow: Challenges States Face in Forecasting Long-Term Expenditures,” was published earlier this year.

This interview with the GFRC research team—Deborah A. Carroll, Ph.D., principal investigator; Christelle Khalaf, Ph.D., co-principal investigator; and Thao Pham, Ph.D., research assistant—has been edited for clarity and length.

Q: How does this report add to what we already know about state forecasting?

Carroll: There’s very little research on expenditure forecasting; most of the forecasting research that’s done by academics and practitioners focuses on the revenue side. So this report helps fill that gap by looking at the other side of the equation.

Q: And how did you carry out this work?

Pham: Our research entailed understanding existing literature and what it offered us, selecting states where we wanted to do interviews, and developing an analysis and write-up based on what we learned.

Q: What can you share about the challenges state staffers raised when you spoke with them?

Khalaf: Our interviews always went back to Medicaid. The size of that expenditure category, coupled with the assumptions that analysts have to make about future fiscal and economic conditions to develop the forecasts, creates large challenges.

Carroll: And then there’s also the challenge analysts face in trying to distinguish between one-time versus ongoing spending.

Another challenge the interviewees spoke about was communicating to relevant stakeholders the nature of forecasting—and the spending decisions that result from forecasting. These stakeholders might be users of financial and budgetary information; they also might be the general public. So, it can be pretty challenging to communicate to such a broad range of stakeholders, and to really have them understand the importance of these forecasting processes as well as the decisions that come from the forecasts.

There’s also a lot of concern in state fiscal offices about an upcoming loss of institutional knowledge from pending retirements.

Q: Interesting. Anything else?

Khalaf: Some people we talked to mentioned that their state’s forecasting methods were too simple or that they faced challenges around getting enough data. For example, the interviewee from North Carolina said that they didn’t have any documentation of historical policy changes and the impact on expenditures of those changes. So, when they tried to use certain models that include changes over time, the models didn’t lead to accurate or reasonable forecasts or results.

Q: Is there a particular reason why some forecasting methods aren’t more sophisticated?

Khalaf: Well, it’s because, in some of the states where we interviewed staff, the use of certain methodologies is required by statute.

Q: Did you find any of this surprising?

Carroll: Yes. I was surprised that the techniques used for long-term expenditure forecasting are much more simplistic than what’s generally used for revenue forecasting, and I say that as someone whose research tends to focus on the revenue side.

And I was actually a bit surprised by the lack of integration—and by that, I mean integration into long-term expenditure forecasts—of things that we sort of know are coming down the line, such as decarbonization, climate change adaptation, and other larger problems that we seem to be talking a lot about. Those things will certainly impact different states differently. 

Khalaf: To me, the reason that emerging risks aren’t integrated into long-term expenditure forecasts in some states is because the methods required by statute are very rigid.

Q: But you could argue that the impact of these emerging risks can be difficult to fully assess because they’re likely to extend beyond even a three- or five-year forecast period. Did that come up in your interviews?

Khalaf: Yes. In Louisiana, for example, the interviewee said that whenever they’re dealing with a big expense they know is coming but is beyond their five-year forecast period, they list it as an “on-the-horizon” item.

Carroll: I agree that a lot of these emerging risks are much longer term and so they may not be appropriately placed in the time frame we looked at. At the same time, a lot of these risks are present right now. If we look at the prevalence of hurricanes and tornadoes and storms in general: a lot of states are already being impacted by these events, which are projected to worsen over time.

So, trying to plan for a full transition to electric vehicles? Yes, that’s probably much longer into the future. But trying to plan for and address some of these more immediate situations, such as the weather events I just mentioned, should be given attention right now.

Q: What might help states overcome some of the challenges you found in producing long-term expenditure forecasts?

Carroll: It would be really helpful for them to have a place where they can access information about best practices and what outcomes result from these practices. For example, because forecasting for Medicaid is something that came up frequently as a challenge, it could help to have a web page that states could search to learn how other states approach their long-term forecasts for Medicaid—and how accurate these forecasts are in hindsight.

In addition to that, technical training to educate stakeholders, legislators, and the press more broadly about some of these concepts could help states better communicate their processes and decision outcomes.

Q: Any final thoughts or insights?

Khalaf: Something that was interesting to me was that people in all the states we interviewed wanted to know how other states do their forecasts and what assumptions other states are using. There’s definitely an appetite for that. And to me, that’s indicative of states wanting to collaborate with each other, or at least share information with each other.

Carroll: They also want to have benchmarks from other states to assess themselves on.