David W. Young

A hospital that does not have an activity-based costing system could be putting its financial viability at considerable risk.

At a Glance

Two measures for computing the cost of intermediate projects-a ratio of cost to charges and relative value units-are highly flawed and can have serious financial implications for the hospitals that use them. Full-cost accounting, using the principles of activity-based costing, enables hospitals to measure their costs more accurately, both for competitive bidding purposes and to manage them more effectively.

For decades, hospitals have used one of two methods to compute the cost of their intermediate products (such as a laboratory tests and radiological procedures): a ratio of costs to charges and relative value units.

Both methods are highly flawed. CFOs who use either of these methods to ascertain the full cost of an intermediate product are making a mistake that can have serious financial consequences for their institutions.

Indeed, CFOs who use one of these methods during negotiations with payers about capitation or subcapitation rates, diagnosis-related group payments, or fees for individual services may end up proposing prices that are too high to obtain the contract. Or perhaps worse, the hospital might end up with a contract in which its service-delivery costs exceed the associated revenue.

The Stage-Two Dilemma in Full-Cost Accounting

A full-cost accounting effort goes through two stages. In stage one, the hospital's accounting staff:

  • Defines cost centers, distinguishing between production (or revenue) centers and service centers
  • Assigns all costs to one or more of those cost centers
  • Determines appropriate bases of allocation for service centers
  • Allocates the service center costs (sometimes called "overhead" or "indirect costs") to the production centers

The end result is that all costs reside in production centers. Although it has some flaws, the stage-one effort is well developed in many hospitals and, when done well, results in reasonably accurate production-center costs.

It is in stage two, when a production center's costs are attached to its products, that problems arise. During this stage, the production center's costs are divided into direct and indirect categories. Direct costs are those that can be attached to a product unambiguously; they typically comprise direct labor and direct materials. There is no big problem here.

With indirect production center costs, however, there are some tricky accounting problems, in part because these costs fall into three categories which, in a more general (or manufacturing) sense, are grouped under the label of "manufacturing overhead":

  • Indirect labor, such as supervisory time
  • Indirect materials (e.g., materials that cannot be directly associated with a unit of output, such as cleaning solvents for machines)
  • Service center costs that were allocated to the production center during stage one

In a typical manufacturing context, MOH ordinarily is "attached" to products through the use of one or more "overhead rates." The absorption process, as this effort sometimes is called, can be a little tricky and can easily give misleading results. For example, when only one overhead rate is used, as happens in some manufacturing settings, the implicit assumption is that the unit used in that rate (e.g., machine hours or direct labor hours) drives the use of all MOH. However, MOH generally results from a more complex array of forces, such that an absorption process using a single overhead rate can give management incorrect information about the actual consumption of MOH, and hence about the full cost of a product. This can lead to poorly informed pricing decisions and profitability analyses.

Enter Activity-Based Costing

Activity-based costing, or ABC, is now used in many manufacturing settings as a way to correct for this deficiency. Designers of ABC systems use multiple "overhead cost pools" and try to make the resources in each pool as homogeneous as possible. Then they identify an activity that drives the use of each pool's resources and use it to compute the pool's overhead rate. For example, one MOH cost pool in a factory might be the labor and supervisory time needed to set up the machines for manufacturing a batch of products. In this case, the appropriate cost pool would be everything associated with setting up the machines (such as cleaning and adjusting tolerances), and the appropriate unit of activity for the pool would be a setup. As a result, a unit of output in a small batch of products would get a higher share of the setup cost than a unit in a large batch.

Unfortunately, although many hospitals have developed intermediate-product costing systems in an attempt to determine the full cost of a DRG (the final product), few have developed an ABC system. Yet intermediate products in a hospital-such as laboratory tests, radiological procedures, and surgical interventions-are, from a cost accounting perspective, conceptually identical to products in a manufacturing setting. Instead of developing an ABC system to cost out the products, however, these hospitals have relied on either RCCs or RVUs as a means to compute the full cost of their intermediate products. The simplified full cost report for Owen Hospital (a fictitious institution), shown in Exhibit 1, can be useful in illustrating the errors associated with each of these approaches.

View Exhibit 1


Cost centers that produce a single product do not need an ABC system. For example, the renal dialysis department at Owen Hospital had direct costs of $1,250,000 and allocated service center costs of $423,930, resulting in a full cost total of $1,673,930. If, during the year that it incurred these costs, the department performed 6,000 dialysis procedures, the full cost of a single procedure (the intermediate product) would be $278.99. In this situation, since the department's only product is a dialysis procedure, and since all dialysis procedures are more or less identical, the average cost per procedure is a meaningful number.

By contrast, the radiology department (direct costs of $1,750,000, allocated service center costs of $688,321, and total costs of $2,438,321) conducts a wide variety of procedures, such as chest X-rays, limb X-rays, CT scans, and magnetic resonance imaging. Each procedure requires some technician time, and some procedures require supplies, such as a contrast medium. These are all direct costs of the department and are contained in the $1,750,000 figure. They are also direct costs of the procedures. That is, it is relatively easy to use time and motion studies to determine the portion of the cost for any given procedure that results from technician time, films, contrast media, and other supplies.

But what about the supervisor in the department who doesn't work on procedures? Or the department's scheduling personnel? Or any of a variety of other people in the department who don't work directly on procedures? These are all direct costs of the radiology department-they can be unambiguously associated with the department, but they are indirect with regard to any given procedure. And yet, if we are to know the full cost of a particular procedure, we must find a way to attach some portion of these costs to it.

Then there are the $658,321 in service center costs that were allocated to the department during stage one. They are indirect with regard to both the department and any given procedure, but we also must find a way to attach a portion of these costs to each procedure. Historically, this is where RCCs and RVUs came into play, and unfortunately, their use has led to potentially serious inaccuracies.

The Flaws in RCCs

To compute the RCC for, say, the radiology department, the accounting staff divides the sum of the department's full costs for a year, as determined in stage one, by the department's total charges for that year. The resulting ratio can be multiplied by the charge for any given procedure, such as an MRI, to determine the procedure's cost.

Research has indicated that the RCC approach to determining a hospital's costs is about 95 percent accurate at the product line level and about 85 percent accurate for a DRG. (For details, see Michael Shwartz, David W. Young, and Richard Seigrist, "The Ratio of Costs to Charges: How Good a Basis for Estimating Costs," Inquiry, Fall 1995.) Below that level of aggregation, however, its accuracy declines, and it is extremely unreliable for any single product provided by a production center.

View Exhibit 2


To illustrate this inaccuracy, assume we wish to compare the costs of four laboratory tests used in the treatment of a patient with a given DRG in a teaching hospital with the costs of the same four tests for a patient with the same DRG in a community hospital. Assume the following conditions apply:

  • Each hospital conducts four identical tests-three "simple" tests and one "esoteric" test.
  • The laboratories in the two hospitals are equally efficient and pay the same factor prices (wage rates, etc.). Thus, there should be no real cost differences between the two hospitals. That is, the "true cost" of each test (in terms of resources actually expended) is the same in the teaching hospital as the community hospital.
  • For various strategic reasons, however, the two hospitals use different "markups," resulting in different charge structures, but these differences net out at the department level. That is, the average markup is the same in each hospital's laboratory, as shown in the exhibit above.

Thus, although the true cost of each test is the same in both the teaching hospital and the community hospital, the reported costs (using the average RCC of 0.48 multiplied by the charge) are quite different.

View formula 1


As a result, total costs for the four lab tests in question would appear to differ considerably.

View formula 2


In short, for two patients receiving the same four tests, the cost in the community hospital would appear to be $79.20, compared with $108 in the teaching hospital. And yet the actual resources consumed were identical in both institutions.

Over a large volume of tests, these cost differences would net out. But in a department that provides a heterogeneous mix of outputs, the use of RCCs usually will produce similarly misleading results for any aggregation of patients below the DRG level, and certainly for any single patient or any single test or procedure.

It often is argued that teaching hospitals are more expensive than community hospitals because of their greater case complexity, research agendas, residency programs, location, and a variety of other matters, all of which quite likely is true. However, it also seems likely that at least some portion of their so-called higher costs arises from the flaws in an RCC approach to costing. Indeed, imagine the difficulty the teaching hospital in the above example would have in obtaining a managed care contract in which the population to be served was expected to use a large number of "simple tests," and it thought that its cost for a simple test was $24 instead of $10.

The Flaws in RVUs

Many hospital departments perform a variety of different kinds of procedures. In radiology, these range from a simple chest X-ray to a rather sophisticated CT scan that requires injecting a contrast medium into the patient's blood and taking several images of the patient's brain as the contrast medium flows through it in the bloodstream.

Assume that during the period covered by the cost report for Owen Hospital, the radiology department performed 30,000 procedures. With total costs of $2,438,321, the average cost per procedure is $81.28.

Assume further that we wish to know the radiology cost for two patients, each of whom received two procedures. If we use the average cost per procedure, the cost for each patient would be $162.56. However, if Patient A received two chest X-rays and Patient B received two CT scans, these totals would be very misleading.

To address this problem, the radiology department might decide, as many others have, to use relative value units. An RVU approach uses information from a national database to assign a weight to each procedure based on its complexity. For illustrative purposes, assume a chest X-ray is worth one RVU and a CT scan is worth 10, and that the radiology department performed 25,000 chest X-rays and 5,000 CT scans during the period covered by the cost report. We can compute the total RVUs delivered during the period as follows:

View formula 3


Since Patient A had two chest X-rays and Patient B had two CT scans, the new per-patient costs are shown in the exhibit below:

View Exhibit 3


Intuitively, this makes sense, since we know that a CT scan is a more complex procedure than a chest X-ray. Given the assumption that a CT scan is 10 times more complex than a chest X-ray, based on its RVUs, it makes sense that its cost should be 10 times that of the X-ray.

In terms of accuracy, however, the important cost accounting question is not procedure complexity, since much of this is picked up in the material and labor costs that are attached to the product directly, but whether all MOH should be a function of RVUs. If we conclude that MOH is driven by something other than procedure complexity, then we must undertake an ABC effort. To do so, we need to divide our overhead costs into several pools and determine the appropriate cost driver for each. The exhibit located below illustrates how costs might be computed if drivers other than RVUs were used for attaching the department's MOH to its products.

View Exhibit 4


There are several important points to note about this exhibit. Labor and material costs are direct costs and therefore are not an issue. (Because the CT scan has 10 RVUs, its unit costs under the RVU approach are 1/10 of what they are under the procedure approach. In total, they are the same under both approaches.)

There are five overhead cost pools. Four (hospital administration, plantwide depreciation, housekeeping, and maintenance) are the same as the ones used in Exhibit 1; that is, they are hospital service centers whose costs were allocated to radiology, and the total amount shown is the amount of the allocation. The fifth is the department's administrative costs, which are direct costs of the department but indirect with respect to any given procedure performed in the department. (In practice, we might use more overhead cost pools in an effort to improve the cost homogeneity in each.)

To attach the costs in these pools to the department's products, this example uses the same cost drivers that were used to allocate the costs to the department in Stage 1 (such as square feet for depreciation and housekeeping). It then computes the amount of space, maintenance hours, and salary dollars associated with each of the procedures, and uses the resulting percentages to distribute the costs between the two procedures.

The specific results would change depending on the overhead pools we select, the cost drivers we use for them, the mix of procedures, and the number of cost driver units used by each. With the assumptions in Exhibit 4 showing how costs might be computed using ABC, however, the per-patient cost comparisons between RVUs and ABC would look as shown in Exhibit 3.

With the ABC approach, we have relaxed the implicit assumption under the RVU approach that all overhead is driven by procedure complexity.

In its place, we have attached overhead to each product by means of several homogeneous overhead cost pools and using an appropriate cost driver for each. The result is that the chest X-ray now costs about twice as much as under the RVU approach, and the CT scan costs about half as much.

More generally, hospitals that substitute RVUs for a serious cost accounting effort can come to highly misleading conclusions about the cost of an intermediate product. This occurs because, although RVUs address the complexity issue, they do not deal with the fact that much of MOH is unrelated to procedure complexity.

When to Use ABC

Hospitals and other healthcare providers that wish to measure their costs more accurately, both for competitive bidding purposes and to manage them more effectively, can benefit considerably from ABC. Clearly, there are situations where a single overhead rate is adequate (such as in the dialysis unit) and, therefore, there is no need for ABC. But there are many other situations where the presence of one of three factors-product diversity, relative costs, or volume diversity-calls for an ABC system.

Product diversity exists when different products use overhead-related services in different proportions. For example, if one product requires considerably more supervisory time than another, there is product diversity. However, product diversity is important only if the costs of the different activities are significantly different, which is the relative cost factor. Finally, some overhead activities are batch-related, and if products are manufactured in batches of different sizes, there is volume diversity.

Developing an ABC System

The details of the process for developing an ABC system are discussed in many cost accounting textbooks. In general, the process entails classifying a production center's costs into four activity-related categories.

Facility-sustaining. This is the highest-order activity, and includes work such as plant (or departmental) management, repair and maintenance, security, and other activities unrelated to the department's products, per se.

Product-sustaining. These are the activities needed to ensure that products are produced according to specifications. They include process engineering, product specifications, engineering changes, and product enhancements.

Batch-related. These are activities that are performed each time a batch of products is manufactured, such as set-up time for machines, material movements, and inspections.

Unit-level. These are activities that are tied directly to the number of units produced, such as utility usage, machine hours, and the like. (Unit level activities also include direct manufacturing costs; the three other categories include only manufacturing overhead.)

These four categories are the building blocks of an ABC system. Once costs have been classified into a category, and the categories have perhaps been further subdivided into more homogeneous cost pools, the accounting staff needs to determine the activity that causes a product to incur the costs in each pool. Conceptually, the search is for a cost driver that reflects a product's demand for the resources in the pool. For some pools this is relatively easy; for others, it can be tricky. Nevertheless, even if considerable effort is required to put an ABC system in place, it is a one-time effort. Once developed, an ABC system can be sustained at minimal ongoing expense.

ABC: Taking the First Step

In summary, a hospital department can use a single rate to assign its MOH to products if it has little product diversity and production batches that are all about the same size, or if it has high product and/or volume diversity but similar activities regardless of the type of product or the batch size in a production run. A dialysis unit, with little product diversity and equal batch sizes (one dialysis in a "batch"), falls into this category.

By contrast, when there is high product or volume diversity, or differences in activities associated with different products or batch sizes-as there are in pathology, radiology, the operating suites, and many other departments where a hospital's intermediate products are delivered-there quite likely is a need for an ABC system with multiple overhead cost pools and a separate cost driver for each. Once such a system is in place, it can produce more accurate cost information, which, in turn, can better inform pricing decisions and profitability analyses, especially in hospitals that are engaged in managed care contracting.

Outside of health care, the use of ABC has led managers in many organizations to reverse their thinking regarding which of their products are the most profitable. It is likely that similar conclusions would be reached in a hospital that undertakes an ABC effort. Indeed, in an era of shrinking resources for health care and intense negotiations between payers and providers, a hospital that does not have an ABC system could be putting its financial viability at considerable risk.

David W. Young, DBA, is professor of management, emeritus, Health Sector Management Program, Boston University School of Management, president and CEO of The Crimson Group, Inc., Cambridge, Mass. , and a member of HFMA's Massachusetts Chapter (DavidYoung204@cs.com).

Publication Date: Sunday, April 01, 2007

Login Required

If you are an existing member, please log in below. Username and password are required.



Forgot User Name?
Forgot Password?

If you are not an HFMA member and would like to access portions of our content for 30 days, please fill out the following.

First Name:

Last Name:


   Become an HFMA member instead