This paper describes work in progress on the Higher Education simulation project funded by the Alfred P. Sloan Foundation. Contents may not be used or cited without permission. Limited distribution is provided to obtain comments and criticisms, and to assist potential development partners. Copyright © 1997 by JHHEG.

Higher Education is a computer-based simulation game under development that targets both the institutional professional and the interested layperson to participate in leadership challenges in a college or university setting. Players set, monitor, and modify a variety of institutional parameters and policies, allocate resources as they see fit, and watch as results continually unfold. The game provides an opportunity to experiment and succeed or fail in a safe and entertaining fantasy environment. While Higher Education is necessarily a caricature of real academic life, it is grounded in authentic data and will provide serious lessons in higher education. The game will be driven by a sophisticated simulation engine that models five broad areas:

The plant activities model deals with the built environment in which Higher Education’s various operations take place. The model consists of four components: space norms and the existing space inventory; plant operations and maintenance (O&M); planning and construction of new facilities; and student residence halls. "Plant" refers to all buildings except those associated with residence halls and associated food service. For simplicity, the residence hall model includes both annual operations and new construction.

The space norms and existing space inventory component translates information about the scale of operations into normal square footage requirements for each academic department and the central administrative and support functions. It also keeps track of the existing space assignments for each unit. The O&M component determines the normal yearly expenditure requirement for the player-generated institution’s existing space inventory and tracks the rise and fall of deferred maintenance. The facilities planning and construction component manages the institution’s capital reserve and plant debt, determines the level of funding for the year’s construction program, and then allocates these funds among alternative construction and conversion projects. (Players may take away a department’s space and convert it for use by another department.) The residence halls component tracks demand for existing residential capacity, floats new residence hall debt and constructs new capacity if warranted by demand and player policy, and sets the room and board rate so that the residence halls operating unit just breaks even.

The physical plant model will run once per simulated year. The residence halls component will run just before the resource allocation model, the other components just after. "Next year" refers to the year for which resource allocation has just been completed. "Base year" refers to the year just being completed.

This model first calculates the normal square footage requirement associated with each department’s operating level during the base year. For purposes of the physical plant model, "department" means the list of academic departments plus one central operating unit which encompasses all institution-wide activities except residence halls. Normal square footage for the academic departments is calculated as follows:

Benchmarking studies show that modern academic facilities require maintenance in the order of 1 1/2 to 2 percent of replacement cost in order to avoid building up a deferred maintenance backlog. To that must be added the cost of utilities, custodial services, and other operating expenses. All these figures vary somewhat by type of facility. Laboratories cost more to maintain and operate than offices., for example. In view of the above, Higher Education specifies a "normal O&M requirement" as a percentage of replacement cost per square foot, for each academic department and the central administrative unit.

The plant O&M model component compares next year’s actual O&M expenditures as determined by the resource allocation model with the normal O&M requirement as determined from next year’s beginning space inventory. Shortfalls add to the deferred maintenance backlog, surpluses reduce it. (The deferred maintenance backlog can never become negative, however.) The deferred maintenance backlog is expressed as a percentage of replacement cost. Deferred maintenance is assumed to be spread evenly among campus buildings, so the percentage does not vary by department.

The deferred maintenance backlog will inform the academic operations and institution-level performance functions.

Operational expansion creates demand for new facilities. Even if faculty, student, and staff counts are constant, depreciation and obsolescence open gaps between the space norms and the available square footage. The facilities planning and construction model determines the funds available for closing these gaps and allocates the funds among alternative construction and conversion projects.

The model begins by adding the resource allocation model’s transfer to plant to the base year’s capital reserve balance to obtain the new beginning balance. Interest on the beginning balance, at the short-term interest rate, also is added at this stage. The resulting sum may be augmented by new debt at the player’s discretion, providing that the institution has not exhausted its debt capacity. Consistent with real-world practice, any new debt issue will have to exceed a threshold amount. (The cost of issuing small increments of long-term debts is prohibitive.) The player then decides how much to spend for projects during the next year and how much to retain to fund future needs.

The player’s spending decision is informed by a report that shows each department’s space shortfall or surplus: i. e., the difference between its actual square footage and the norm. The report also shows the total cost of closing any gaps without converting space for use by different departments. (Whether to allow conversion is up to the player, as discussed below) This figure, which represents the extreme scenario, may well exceed the institution’s financial capacity.

Absent player intervention, the model uses a simple default policy to determine annual spending for capital projects: the remaining capital reserve balance will be set equal to the beginning balance, escalated by construction cost-rise. This is a conservative policy, justified by the assumption that the beginning balance already represents the minimum required to safeguard against future contingencies. No new debt is allowed. The player can override the defaults at any time by setting specific figures for new debt and project funding—in this case the new ending balance becomes the starting point for reapplication of the default policy.

The model optimizes the year’s facilities construction and conversion plan given the available funds and the policy about whether to allow conversion. (The default is "no conversion.") Players may perform sensitivity analyses to test the effects of conversion and different funding levels. New square footage provided for in the final facilities plan will become available to departments one year hence.

The facilities planning model uses linear programming to maximize the weighted sum of the new square footage to be provided for each department, subject to certain constraints. There are two decision variables for each department: incremental square footage from new construction and space obtained by converting unused space currently assigned to a different department. There are two sets of weights. The departmental weights determine the player’s priorities for meeting the various departments’ space needs. For simplicity, they equal the academic department priorities used by the resource model to allocate faculty lines. (Because the central "department" does not have a faculty priority weight, it will be set equal to a multiple (e.g., 90 percent) of the average departmental weight.) The second weighting factor represents the desirability of new construction vs. conversion. Converted space involves an opportunity loss for the source department and it generally is less useful for the receiving department than new construction. Therefore, it receives a lower weight than new construction.

Three constraints apply regardless of the policy on conversions.

The rows represent the source of the space to be converted (the department that will give up space in the reassignment) and the columns represent the department that will receive the space. The entries represent the proportion of each department’s space that can be converted for use by another department. For example, 85 percent of SS1’s available space can be converted and reassigned to SS2. These figures, which reflect technology rather then policy, will be provided as part of the game’s initial conditions.

The model uses linear rather than quadratic programming to limit the number of projects fielded in a given year. The linear program will generally satisfy all the needs of the highest priority department before addressing those of the next highest priority department. Quadratic programming, on the other hand, would produce larger numbers of smaller projects, each satisfying only a part of each needy department’s requirements. The former is more realistic because small projects tend to be inordinately costly. In addition, the player will find it easier to track a smaller number of larger projects.

Following real-world practice, Higher Education treats residence hall and food service operations as an auxiliary enterprise—that is, separate from the main elements of university finance. An auxiliary must cover its costs from its own revenue sources; in this case room and board fees paid by traditional undergraduate students. (Graduate students and non-traditional students are assumed to be ineligible for campus housing.) The cost structure includes expenditures for operations and for servicing any debt incurred to finance dormitory construction. To simplify matters, the residence halls operation (which henceforth will be considered to include food service) is required to break even every year. This means players need not concern themselves with surplus/deficit and reserve levels.

The residence halls model will run in tandem with the resource allocation model. The steps in the algorithm are as follows:

Summing the room and board and tuition rates produces the gross sticker price for traditional undergraduates. (The gross sticker price informs the enrollment management model, discussed in Technical Document 2.1.) Player-initiated adjustments to operating cost affect the quality of student life as well as the room and board rate. Spending more is assumed to improve quality, and better quality may boost the demand for residence hall spaces one year after the extra cost is incurred. In other words, cost adjustments affect price immediately and quality after a year’s lag. Simplicity suggests that the effect of price on demand also should be delayed by a year. (Otherwise the model would have to solve simultaneous equations to get the effect of price on demand and the effect of demand on unit cost and price.) One can argue that students are be slow to react to price changes, in which case the model may in fact capture an important element of reality.

Following the above, the model projects occupancy as a function of base year demand, the base year’s price-quality ratio, and exogenous factors. Because the residence halls operation must break even, actual occupancy always equals projected occupancy. Operating costs are then determined as a function of occupancy and capacity. Some costs vary with occupancy while others vary with the total number of residence hall spaces, whether occupied or not. (The proportions are assumed to remain constant over time, except for the effects of new debt service.) Dividing total cost by the number of occupied spaces yields the room and board rate for the year.

The model will add residence hall capacity if the demand for spaces exceeds supply by a sufficient amount. (The threshold may be modeled as a discreet variable: e.g., "conservative, moderate, aggressive.") Demand depends on the number of traditional undergraduates, the normal fraction who wish to live on campus (an exogenous variable), and any adjustments due to price-quality changes. Excess demand exists when demand exceeds capacity. The model creates new residence hall capacity automatically whenever excess demand exceeds the threshold. All new capacity is assumed to be 100 percent debt financed at an amount equal to the number of spaces times a parameter for capital cost per space. The new spaces will become available one year after the decision to build, and the new debt service will be added to fixed cost at the same time.

The residence hall model requires only two player-controlled policy variables: the amount, if any, by which operating cost is to be adjusted; and the excess demand threshold for creating new capacity. The default for expenditure adjustments is zero (i.e., quality is to remain constant). The default for the capacity policy variable might be "moderate."