This paper was presented at the 1997 CAUSE annual conference and is part of the conference proceedings, "The Information Profession and the Information Professional," published online by CAUSE. The paper content is the intellectual property of the author. Permission to print out copies of this paper is granted provided that the copies are not made or distributed for commercial advantage and the source is acknowledged. To copy or disseminate otherwise, or to republish in any form, print or electronic, requires written permission from the author and CAUSE. For further information, contact CAUSE at 303-449-4430 or send e-mail to info@cause.org.

Regardless of how they are budgeted, IT costs must be paid. Most schools offer some computing services at no cost to the individual or unit and some that are charged back to users and units. Typically, common good services are centrally funded, and services that differentially benefit specific individuals or units are charged for. How services are funded often reflects a school’s philosophy about IT and about finances. Preferably, IT funding mechanisms deliberately help shape and influence an institution’s IT and services philosophy, as opposed to an IT or service philosophy being unintentionally shaped by fiscal policies that follow no particular strategy. Levying fees to users on an individual or departmental basis may yield a different demand and expectation of IT services than when costs are borne by a central budget. Quantity and quality, degree of centralization, and administrative complexity of services are major variables in determining funding. Also at stake is the degree to which an institution wishes to endorse, suppress, control or expand IT services. These issues are specifically illustrated through a case study of the formulation of a new budget and cost accounting model to both finance an institution-wide network upgrade and to maintain that network.

This paper presents a case study of how Dartmouth College pays for its physical data network. We first present a brief overview of paying for IT costs in order to provide context for our case study.

For our purposes, there are three types of IT consumers: faculty, students, and staff, and a funding continuum anchored by individual charges at one end and central charges at the other. A typical example at the individual charge end is the computer store. Generally, an individual as a customer at the store buys goods at a certain price. A typical central charge is basic consulting. You can usually call or visit a consultant to ask a basic question without paying an individual fee. Even these examples may be reversed. Faculty, students, or staff may be regularly provided with a computer from centralized funds (for students this could be “buried” in tuition), and some schools might charge individuals for consulting contacts. Any such examples can get much more complex. Computer stores, whether outsourced or not, may be partially centrally subsidized. The “profit margin” charged at a store may be used to subsidize other IT services, perhaps basic consulting services. Consulting may charge for some services and not others. Also, a service may also be provided “free” to some individuals but at a cost to others.

Unless an institution has strict and limited financial models that affect IT and all other services, there are many possible ways to pay for IT. It would be hard to find a school that is completely at one end or the other of the funding continuum; i.e., a school that devotes no general funds to IT services or that charges for every single IT service it provides at the individual level. And so, the fun begins.

It’s been remarked that a large pile of horse manure left on a busy street corner with a sign that says FREE will be quickly gone. Central funding, typically for some service considered basic and crucial to an institution, can lead service consumers to think a service is “free.” However, only senior administrators will likely truly appreciate the costs of a “free” service. A simple example is plant services. You expect there to be electricity, heat, water, etc. in your work environment. When you turn on the lights at your office you don’t think much about the cost. Is your institution encouraging you to waste electricity, to efficiently depend on a basic and necessary service, or both? In general, the more a service is considered to be critical, the more centrally it is funded. The opposite is also true. If senior administration does not see a service as critical, it receives little central support, either monetarily or otherwise. A single service might illustrate both cases. For example, an institution might offer “free” (100% centrally funded) videotaping of a regular credit-granting course, but, fully charge an individual office for a guest speaker whose presentation is open to the public. Even these examples can become complex. An office that brings in many public speakers might successfully argue for a central budget increase to support the rising costs of recording famous speakers.

As IT professionals, we can consider the variety of funding mechanisms that may be available to us not as impediments to all the service demands we try to fulfill, but, as creative possibilities to achieve sufficient and ongoing funding for our programs.

When a service should be centrally or non centrally funded is a complex issue that depends on an institution’s history and culture, as well as economic factors and the service in question. In general, a common good service, one whose base levels should be equally available to have and have-not units, is better centrally funded. Services that are not a necessity for each unit, or at least services beyond a minimal base level, might be charged back at the unit or individual level. Centrally funded services can be more cost-effective than non centrally funded services, or they can be relatively cumbersome and unresponsive. This latter instance may seem especially true to departments that have more resources than others. There’s no simple answer and solutions are typically situational.

The funding of an IT service may be altered to be more or less centralized. Moving from central to non central funding may be seen by some as unfair cost shifting. A centrally funded service at least has the appearance to individuals of being “free.” If central funding is decreased or removed and chargeback begins, then something that was at least “free” suddenly must be paid for by individual or departmental funds, even if absolute costs to the institution decrease. Arguments for moving from central to non central funding are typically to free up central funding for a more crucial service and to let “free market forces” determine the fate of a less critical service. Moving a service from non central to central funding might occur in two ways. The rarest way is for new central funds to be allocated with no concomitant recovery of funds from non central units to a central budget. More typically, a non centrally funded service’s funding is centralized by identifying funds allocated to this service in budgets around an institution and rolling those funds into a central pool. As one might guess, to be successful, this transition is not typically done by fiat but through a consensus building process and a supportive fiscal administration.

At Dartmouth, our data network has been a source of technical pride since 1985 when it became campus-wide, including the residence halls. Network funding, however, has had a more tenuous history. Effective FY 1996, we took a mixed central and non central funding model that had inadequate support for network maintenance and no support for network upgrades, and created a central model that dynamically provides for both. Briefly, network funding had been based for over ten years on a hodgepodge of charged back and central funds. The chargeback mechanisms had decayed to the point where they yielded a relatively small amount of funds and had little logic to them. Central funds were frozen at a fixed amount. This situation became intolerable when it became obvious that most of the physical network had to be upgraded, a project with estimated capital costs of $4.5 million.

Driving the decision to alter the funding model was not only recognition by senior officers that an upgrade was necessary, but that the data network was the sine qua non of Dartmouth’s IT environment, that the IT environment was critical to scholarship and administration, and that therefore, a data network was a common good that should be centrally provided at a minimal base level to all segments of the institution, without individuals or non central units and departments needing to budget for it, at least at a base level. This recognition was accomplished by a series of communications from Computing Services to key senior administrators, and included working with one or two such administrators who demonstrated a willingness to champion the cause of the network with other decision makers.

Upon the encouragement of the Vice President for Finance, Computing Services also actively partnered with plant operations (at Dartmouth, known as Facilities Operations and Management) in all phases of the network upgrade project. This partnership resulted in even greater project momentum, and provided a beneficial exchange of project management techniques. Facilities Operations and Management provided expertise regarding the installation of conduit, construction, and finish work. The exchange of project management techniques was at times a source of great conflict, but has finally resulted in a comfortable and productive ongoing relationship that continues to generate subsequent projects. This partnership also resulted in the network upgrade project budget including much needed upgrades for phone and video, and support for investment in a within-building raceway and conduit plant that will provide for relatively easy and inexpensive future wiring upgrades.

In short, the data network came to be seen in part as a basic utility, like water, electricity, and heat. At the same time, it remained as the foundation of IT services, a good that is developed and nurtured by IT professionals who form an integrated organization with other IT service providers who provide services layered on the data network — academic and administrative specific applications, and basic consulting support. Funding for the data network remains separate from other services. Data, telephone, and video network funding are three separate budgets. In future years, when technology drives data, video, and telephony together technically, funding mechanisms will likely follow suit.

At present, the data network, both operational maintenance and upgrade capacity, is centrally funded at the level of Dartmouth’s three professional schools, the Office of Residential Life, which is responsible for all student residence facilities, a handful of other auxiliary operations, and Facilities Operations and Management, which has responsibility for the bulk of Dartmouth’s facilities.

The recovery of costs associated with running our campus data network has undergone a significant shift within the last two years. In the past, the various departments on our campus were charged a fixed fee loosely tied to the number of network connections in use by their department. This fee was not supported by any true cost accounting, but, rather, was fashioned (both in amount and the way it was recovered) after the monthly telephone line charge.

Departments budgeted the network port charges as line-items within their program budgets. As with any expense, there was constant pressure to reduce this line-item to free up resources for other programmatic needs. Departments avoided increases in costs by daisy-chaining off of existing network connections, creating not only network management nightmares but less than ideal working conditions for its employees. With hundreds of individual departmental accounts being affected by the network port charges, there was an unmanageable amount of contact between the network billing administrator and the cadre of administrative assistants on campus. Plus, because there was no sound rationale for the charge beyond the need for the IT budget to defray some of the data network operating costs, much discretion was exercised in reducing a department’s charge.

The network operating costs shifted to the departments via this earlier model covered only a portion of these costs borne by the Computing Services organization. Without a mechanism in place to tie the recovery of funds to the actual costs of the network, it was clear a new model was needed as our campus looked at the prospect of upgrading our network, and as the ongoing maintenance of the network put more pressure on the IT operating budget.

At the time support for a network upgrade was being gathered among senior administrators, the College instituted policies governing cost accounting standards and institutional chargebacks. Sparked by federal grants management issues, the College examined a number of areas where costs for services were distributed to client departments; this examination led to requirements for a formal rate calculation to support the costs distributed to other departments. The data network costing model was one of the first services to be assessed and redeveloped under these new College policies. There was no question among the Budget Office, the IT organization and key fiscal administrators that the model for network port charges needed to be revised.

Our emerging partnership with Facilities Operations and Management in planning the network upgrade project led to a natural suggestion that active network connections be considered a facilities expense. As technology continues to become more and more an integrated part of an employee’s experience, it seems obvious to begin thinking of an active network port as a utility; that is, just as one’s office has light, heat, and telephones, the office should have an adequate network connection.

Distributing network costs as facilities costs had a lot of immediate advantages for our campus:

• The facility and its use, instead of the financial conditions of individual departments, begin to define the level of network connectivity.

• Points of contact for questions regarding network cost distribution are reduced to a handful of individuals who have a thorough understanding of the model, resulting in a savings of IT administrative effort.

• An annual cost accounting must be performed and rates adjusted to reflect changes in costs and/or changes in numbers of active ports. This cost accounting must be done and results provided to facilities budget managers in time for setting the upcoming year’s budget. Previously, the model was too loose, had lots of grandfathered deals, and there was no defensible rationale for the amount of the fee.

• With a $4.5 million network upgrade on the boards, we were able to sell the impact to the operating budget as an enhancement of our facilities infrastructure — this ‘sell’ would have been extremely difficult if we had to effect each of the individual operating budgets across campus. Many departments driven by other demands on their budget would have climbed on the ‘we don’t need Ethernet - we’re doing fine’ bandwagon and protested any increase in fees.

After building consensus among the senior fiscal administrators to support a facilities-based network cost accounting model, we worked with the Budget Office to explore how we would be able to seed the FY 1996 facilities budgets with funds to cover these charges. The Budget Office tapped College departments for their individual line-item budgets for network port charges and reallocated these funds to Facilities Operations and Management and the Office of Residential Life. The professional schools and the remaining auxiliary operations were informed of the shift to a facilities-based model and were responsible for their area’s budget reallocation. This shift was seen as an overall positive by the campus as it reduced the noise in the individual departmental budgets, the departments would be getting Ethernet connectivity with the upcoming network upgrade, and there wouldn’t be the same incentive to cut corners on providing adequate network connectivity in a department.

Certain costs of running the network are captured in a discrete general ledger account. These are costs that are easily identified as being dedicated to the operation and maintenance of the campus network: salaries and benefits of network technicians; telephone lines and other connection fees; various hardware expenditures. There are, however, other costs across the organization to maintain the network. These include: systems assurance personnel; administrative support; software development costs. During the transition towards a full cost accounting and facilities-based model, Computing Services was quite aggressive in identifying costs across the organization that could be directly tied to the maintenance of the network. Of course, it was to Computing Services’ benefit to off-load costs to the network wherever possible, but we were held to only those we could make a legitimate argument for to the College’s Cost Accountant (whose interpretation of some items was not nearly as liberal as ours!)

Accounting for the debt service component of the network upgrade project has been more complex. These costs are being funded with external debt, and funds to cover the related debt service will be recovered through the operating budget via the network cost accounting model. Although the project will not be complete until June 1998, we have already begun recovering a portion of the debt service in the FY 1998 network charges. Facilities Operations and Management and the Office of Residential Life will be receiving an additional budget increment to offset the increase in the network port charges from the debt service for the network upgrade.

There is the expectation that the network cost accounting model will break-even on an annual basis. Of course, no matter how detailed our projections are in the spring for the upcoming fiscal year, there will be some under or over-recovery of costs through the network cost accounting model. These costs are reflected in an adjustment to the future budgets and are recovered through or go to subsidize future rate calculations.

See Table 1 for a detailed illustration of Dartmouth’s FY 1998 network cost accounting.

The operating costs of College buildings are managed by Facilities Operations and Management, the operating costs of the student residence halls are managed by the Office of Residential Life, the professional schools are responsible for the management of their facilities costs, and there are a number of auxiliary operations responsible for their own facilities costs. The costs for operating the campus data network are allocated to a particular facilities cost center based upon the number of active ports in a given facility. The Technical Services group within our IT organization maintains a database of all active network ports on campus down to the room location. This information is summarized by building and feeds into a model that aggregates the number of active network ports by affected cost centers. The calculation for allocating costs to these cost centers is straightforward:

[# Ports Assigned to a Facilities Cost Center

÷ Total # Network Ports] x [Total Network Costs] =

Costs Charged to a Particular Facilities Cost Center

The rates for the following fiscal year are set in March so that the affected cost centers can factor any changes into their budget development.

There has been widespread support of allocating charges to the facilities on this basis as there is an intuitive nod that a facility with more active network connections should bear a greater proportion of the network operating costs than a facility with less active network connections. Certain costs of the network upgrade project, such as raceway construction and in-building wiring, are not allocated on a per port basis. In discussions with the fiscal managers of the various affected facilities budgets, we were reminded that some areas (primarily the professional schools) had funded some improvements to their networks in earlier years. These managers were adamant that they not be penalized for earlier renovations and that they not shoulder the cost of others getting to where they have already paid to be. These costs of the network project are allocated to the responsible cost center on a by-building basis. The remaining costs of the network upgrade project (the fiber backbone and the electronics) will be shared by all facilities cost centers based on proportion of active ports.

As the network upgrade project has progressed, the absolute number of ports has grown significantly. Although growth in ports alone will drive the per port charge down, the costs associated with an expanded, upgraded network have also risen. This combination resulted in a 16% increase from FY 1997 to FY 1998 in our network port charges. As context, the per port charge increased only 2% from FY 1996 (the first year this model was implemented) to FY 1997.

By and large, instituting a facilities-based model to fund the operating and upgrade costs for the campus data network has been advantageous to our IT organization. However, some issues have arisen:

• Service demands of individual departments, such as moving port locations, must be consistently managed so that extraordinary demands do not add unreasonably to the costs borne by all citizens of the network.

• One of our professional schools has embarked on a significant technology initiative adding many, many network connections. Because of the high density and sheer volume of port additions, this area did not want to pay the full cost per port given the economies of scale of a dense port installation. Negotiations hinged not on a reduced fee per port but on the number of ports that would be actively used; these negotiations will no doubt be revisited frequently.

• The need to closely manage costs to smooth budgetary impact; IT can’t pass along spikes in cost to facilities and auxiliary budgets without a detailed explanation of what is driving the change in rates. This level of detail was not needed in the previous model as the impact was diluted over many, many departments, and the issue was insignificant relative to a department’s overall budget.

• Certain Computing Services functions (including portions of FTE) are now formally funded by cross-charges coming in from the network model. If the model changes, it may significantly impact our ability to operate other Computing Services functions at their existing level.

We have taken a poorly functioning network funding model, and — driven by upgrade needs and new institution-wide cost-accounting procedures — we have created a centrally funded facilities based model. This model has further developed our relationship with our plant organization. Plant operations is not typically a close partner of a computing organization, but this alliance has proved beneficial in many areas. Development of the funding model has also strengthened our relationship with various financial offices and fiscal officers around the institution, as we worked with a few central areas to develop the fiscal details. The new model, though a vast improvement over the past, is far from perfect. However, we now have a mechanism that has forged new relationships for our IT organization, that meets institutional and federal accounting requirements, and that can be adjusted by consensus and in the light of day with a few central units.

Most importantly, base level data networking is still “free” to individuals and departments, network maintenance costs are tied to a logical and dynamic scheme, provisions have been made to ratchet-up base level network performance as technology and needs advance, and our institution’s senior officers have been brought to affirm the necessary and central role of the data network and of IT for scholarship and administration.