Minimising Financial Risk under Enrolment Uncertainty: An Open-Source Entropy-Indexed POMDP Tool for University Governance

Universities are businesses of a peculiar kind, where the money needed to pay for next year's staff and buildings must be spent years before the students who will fund them even arrive. This creates a constant tension between the need to plan ahead and the impossibility of knowing the future with certainty. In the United Kingdom, this uncertainty has become acute. For years, universities have relied on forecasts to guide their spending, but regulators have found these predictions to be persistently too optimistic, leaving nearly half of English providers facing financial shortfalls. The core problem is not a lack of effort or intelligence among university leaders; it is a structural flaw in how they make decisions. When the future is genuinely unclear, asking a governing board to agree on a single "best guess" scenario is asking them to do something they cannot reliably do. They are asked to assign weights to different futures, but the very nature of the current environment means those weights are unknown.

To solve this, researchers at the University of Bedfordshire have developed a new kind of planning tool that changes the question. Instead of asking leaders to guess which future is most likely, the tool measures how confident the institution can be about any future at all. It uses a method that tracks the flow of student numbers to update a probability map of the university's financial health. This map does not settle on a single outcome; instead, it constantly shifts as new data arrives, showing whether the university is in a period of growth, stability, or trouble. Crucially, the tool calculates a score of uncertainty based on how spread out these probabilities are. When the data is confusing and the probabilities are evenly split, the tool signals that the situation is too unclear to make big, permanent changes. When the data becomes clear and one outcome dominates, the tool signals that it is safe to act. This approach allows universities to navigate deep uncertainty without forcing their leaders to make impossible guesses.

The researchers built this system as an open-source software application, meaning it is free for anyone to use and inspect. The system was calibrated on a decade of open HESA data from four Russell Group institutions, each with a different recent history. Some of these institutions faced sharp drops in international student numbers due to changes in visa policies, while others experienced steady growth. The team then tested the tool against six other common planning strategies by running a Monte Carlo evaluation across 500 simulated enrolment futures. The results showed that the new tool, which they call an entropy-indexed system, consistently produced better financial outcomes than the traditional methods. In these simulations, it achieved the highest average financial reward while keeping the chance of a financial crisis lower than any of the other strategies. The key to this success was its ability to hold back on irreversible commitments, such as hiring permanent staff or building new campuses, whenever the uncertainty score was high.

The tool operates on a simple but powerful logic. It treats the university's financial situation as a hidden state that can be in one of three modes: a low-growth mode where money is tight, a stable mode where things are normal, or a high-growth mode where there is room to invest. The university cannot see these modes directly; it only sees the annual number of students enrolling. The software uses a mathematical process to update its belief about which mode the university is in every time a new enrollment number comes in. If the numbers are confusing, the software's belief remains spread out across all three modes. This state of confusion is measured by a value called entropy. When entropy is high, the tool automatically blocks the university from making big, hard-to-reverse decisions. It forces the institution to wait and gather more evidence. Once the enrollment numbers become consistent and the belief concentrates on a single mode, the entropy score drops, and the tool gives the green light to invest or expand.

This approach offers a distinct advantage over the standard ways universities currently plan. Traditional scenario planning asks leaders to imagine a few different futures, like a best-case and a worst-case, and then try to choose between them. However, this method fails when the leaders cannot agree on how likely each scenario is. The new tool removes the need for that agreement. It does not ask the board to decide which future is most probable; it simply reports what the data says about the current level of confidence. If the data is ambiguous, the tool recommends a cautious posture, such as maintaining current staff levels or focusing on recruiting students to learn more about the market. If the data is clear, it recommends decisive action. This allows the university to be conservative when it is truly necessary and aggressive when the evidence supports it, without relying on gut feelings or political pressure.

The researchers tested the system on four specific types of university trajectories to see how it would handle different real-world challenges. In one scenario, which mimicked the sharp drop in international students seen after 2022, the tool correctly recommended holding back on spending for the first two years while the situation remained unclear. Other strategies that ignored the uncertainty continued to recommend investing during this period, which led to financial trouble in the simulations. Once the data clarified that the university was stabilizing, the tool switched its recommendation to a more active stance, allowing the university to recover faster than those that had remained overly cautious. In another scenario where growth was steady and predictable, the tool eventually recommended aggressive investment, but only after enough time had passed to confirm that the growth was a lasting trend rather than a temporary spike. This prevented the mistake of building new capacity too early, only to find the growth stalled.

A unique feature of this tool is how it presents its findings to university leaders. It generates a two-page document, called a Board Pack, that summarizes the entire analysis in plain language. This document includes a chart showing how the recommended strategy compares to six other common approaches, along with a clear explanation of why a specific decision was made. For example, it might explain that the tool recommended waiting to hire new staff because the uncertainty score was still too high to justify the risk. Because the document is generated by a computer program that follows strict rules, it is fully reproducible. If another leader runs the same numbers through the tool, they will get the exact same result. This creates a clear audit trail, showing that the decision was based on a transparent analysis of the data rather than a subjective opinion.

The researchers also designed the tool to be accessible to people without a background in mathematics or computer science. A university finance director can use the tool in a matter of minutes to get a basic recommendation, or spend an hour calibrating it with their own institution's specific data. The software is available online for free, and it can be run on a standard web browser without needing to install complex software. For institutions that want to keep their data private, there is a version that can be run on a local computer. The tool uses data from the Higher Education Statistics Agency, which is the official source for university enrollment figures in the UK, but the researchers have made the underlying code and the pre-calculated settings available so that anyone can verify the results.

The study highlights that the value of this approach is highest when the future is most uncertain. In times of stability, traditional planning methods often work well enough. But when a university faces a sudden shock, such as a change in visa laws or a demographic shift, the ability to measure uncertainty becomes critical. The tool does not promise to predict the future; instead, it provides a way to manage the risk of being wrong. By automatically constraining decisions when the evidence is weak, it protects the university from making costly mistakes. By releasing those constraints when the evidence becomes strong, it allows the university to seize opportunities that other, more rigid planning methods might miss.

This work represents a shift in how operational research, a field that uses mathematical methods to help organizations make better decisions, is applied to higher education. For a long time, these methods were used for tactical problems, like scheduling classes or allocating resources within a department. This tool moves that power up to the strategic level, helping the highest levels of university governance navigate financial risk. The researchers emphasize that the tool is not a replacement for human judgment, but a support system that makes that judgment more reliable. It turns the abstract concept of uncertainty into a concrete number that leaders can see and act upon.

The implications of this work extend beyond the United Kingdom. Universities in other countries, such as Australia, face similar challenges with volatile international student markets and policy changes. The researchers suggest that the tool can be adapted for use anywhere, provided the local data is used to calibrate the system. The core logic of measuring uncertainty and adjusting decisions accordingly is universal. By making the tool open-source, the researchers hope that other institutions will test it, refine it, and adapt it to their own contexts. This collaborative approach ensures that the tool evolves based on real-world experience rather than remaining a theoretical exercise.

In the end, the paper presents a practical solution to a persistent problem. It offers a way for universities to plan for the future without pretending to know the future. By focusing on what can be measured—the level of confidence in the data—it provides a stable foundation for decision-making in an unstable world. The tool does not eliminate risk, but it gives leaders a way to understand and manage it with greater precision. For a sector that has struggled with over-optimistic forecasts and financial deficits, this kind of clarity is not just a technical improvement; it is a vital step toward long-term sustainability. The researchers invite finance directors and planning teams to try the tool, to see how it performs with their own data, and to contribute to its development. The goal is to create a shared standard for evidence-based planning that can help universities navigate the uncertainties of the coming decades.