Activity-Based Systems:Measuring the Costs of Resource Usage

Robin Cooper and Robert S. Kaplan

Robin Cooper is a Professor at the Claremont Graduate School and Robert S. Kaplan isa Professor at the Harvard Business School.

This paper describes the conceptual basisfor the design and use of newly emerging ac-tivity-based cost (ABC) systems. TVaditionalcost systems use volume-driven allocationbases, such as direct labor dollars, machinehours, and sales dollars, to assign organiza-tional expenses to individual products andcustomers. But many ofthe resource demandsby individual products and customers are notproportional to the volume of units producedor sold. Thus, conventional systems do notmeasure accurately the costs of resources usedto design and produce products and to sell anddeliver them to customers. Companies, includ-ing those with excellent traditional cost sys-tems,^ have developed activity-based cost sys-tems so that they can directly link the costsof performing organizational activities to theproducts and customers for which these ac-tivities are performed.

I. ABC SYSTEMS AS RESOURCEUSAGE MODELS

Activity-based cost systems estimate thecost of resources used in organizational pro-cesses to produce outputs.^ Many people haveattempted to interpret activity-based costsusing their more familiar fixed versus vari-able cost framework, an interpretation incon-sistent with an ABC system's measurementsof resource usage costs. The conventional fixedversus variable cost cleissification arises froman attempt to classify the likely change inspending or supply of a resoxirce. The measiire-ment of unused capacity provides the criticallink between the costs of reso\ux;es used, asmeasured by an ABC model, and the costs ofresources supplied or available, as reportedby the organization's periodic financial state-

ments.^ The following equation, defined foreach major activity performed by theorganization's resources, formalizes this rela-tionship:

Activity Availability = Activity Usage+ Unused Capacity

A simple example illustrates the differencebetween the cost of resovirces supplied and thecost of resources used to perform activities.

y versions of the transactional demand for re-sources appeared in J. Miller and T. Vollman, "TheHidden Factoiy," Harvard Business Review (Septem-ber-October 1985), 142-150, and Robin Cooper andRobert S. Kaplan, "How Cost Accounting Systemati-cally Distorts Product Costs," Management Account-ing (April 1988), pp. 20-27. A more comprehensive ex-planation of the impact of diversity and complexityon indirect costs was presented in the series (k Jour-nal dfCoat Management articles by Roltnn Cooper, "TheRise of Activity-Based Cost Systems: Parts I-IV" (Sum-mer 1988, Pall 1988, Winter 1989, and Spring 1989).

^See, for example, Robert S. Kaplan, "John Deere Com-ponent Works (A) and (B), HBS Cases # 9-187-107 and-108; Robin Coc^r and Karen H. Wnic^i, "Siemens:Electric Motor Works (A)," HBS Case # 9-189-089.

^ e will use the term "outputs" to refer generically toproducts, services, customers, projects, fadlides or anyobject tlmt creates a demand for <nr benefits from or-ganizational activities. Activity-based cost stems as-agn tiie organization's operating expenses to outputsbased on the activities performed for these outputs.

' We have adqpted the terminology of unused capacity,as suggested by Alan Vercio of Texas Instruments,rather than our initial term of "excess capacity." Notall "unused" capacity represents "excess" capacity.

Many helpful and constructive comments on a previousdraft were made by our colleagues including AnthonyAtkinson, Tbshiro Hiromoto, Jean-Francds Manzoni,Falconer Mitchell, Eric Noreen, Krishna Palepu, Wil-liam Rotch, Keith l^lliams, and, especially, G. PeterWlson.

Accounting Horizons/September 1992

Consider a purchasing department in whichthe equivalent of 10 full-time people [the re-source supplied] are committed to processingpurchase orders [the activity performed]. If themonthly cost of a full-time employee is$2,500,^ the monthly cost of the activity, "Pro-cess Purchase Orders," equals $25,000. As-sume that each employee, working at practi-cal capacity, can process 125 purchase ordersper month, leading to an estimated cost of $20for processing each purchase order.® Thus, theorganization, each month, spends $25,000.This expenditure provides a capability to pro-cess up to 1,250 purchase orders [the activityavailability] during the month. During anyparticular month, the department may beasked to process fewer purchase orders, sayonly 1,O(X). At an estimated cost of $20 /pur-chase order, the ABC system would assign$20,000 of expenses to the parts and materi-als ordered by the purchasing department thatmonth. The remaining $5,000 of monthly op-erating expenses represents the cost of unusedcapacity in the purchase order processing ac-tivity.

This example shows why companies needtwo different reporting systems. The periodicfinancial statements provide information onthe cost of activities supplied each period (the$25,000 monthly expense in the purchasingdepartment); and the activity-based cost sys-tem provides information on the quantity(1,000 purchase orders) and the estimated cost($20,000) of activities actually used in a pe-riod. The difference ($5,000) between the costof activities supplied ($25,000) and the cost ofactivities used ($20,000) equals the cost ofunused capacity (or capacity shortage) duringthe period. And this difference is measured foreach organizational activity, defined by theABC system.''

The two systems provide different typesof information for management. The cost ofresources supplied is relevant for predictingnear-term spending. Spending on many orga-nizational resources will not vary with short-term fluctuations in activity volume and mix.That is why these costs have been classifiedas "fixed" in numerous accounting systemsand textbooks.

But measuring and managing the operat-ing expenses of most organizational resovircesas fixed in the short-run does not give muchinsight as to why the resources were acquired,what the resources are currently being usedfor, and the level of resources that will likelybe required in the future. While the cost ofsupplying the resources may be fixed in theshort-run,^ the quantity of these resourcesused each period fluctuates based on activi-ties performed for the outputs produced. Ac-tivity-based systems measure the cost of us-ing these resources, even though the cost ofsupplying them will not vary, in the short run,with usage.

The ABC resource usage cost informationcan be used by managers to monitor and pre-dict the changes in demands for activities asa function of changes in output volume andmix, process changes and improvements, in-troduction of new technology, and changes inproduct and process design. As such changesare contemplated, managers can predictwhere either shortages or excesses of capac-ity will occur. The managers can then eithermodify their decisions so that activity demandwill be brought into balance with activity sup-ply, or they can change the level of activitiesto be supplied in forthcoming periods.

For example, if newly designed customproducts, with many unique parts and mate-rials, are added to the mix, managers mayforecast a much higher demand for the pur-chasing activity, perhaps now requiring that2,000 purchase orders a month be processed.With no change in the process or efficiency ofthe processing purchasing order activity, thisincrease in demand will exceed available sup-

his cost includes the costs of fringe benefits, secre-tarial and administrative support, equipment costs,and space charges associated with each purchaMngdepartment employee.

^Note that this calculation does not use actual activ-ity levels during the period; the denominator repre-sents service capacity not actual usage of this capac-ity.^Later in the paper, we will show how to develop a newformat for the periodic income or expense statementthat highlights the costs of resources used and unused.^More accurately, the spending on (or expenses as-signed to) these resources will be independent of thevolume and mix of outputs produced during the pe-riod.

Activity-Based Systems: Measuring the Costs of Resource Usage

ply by 750 purchase orders per month, a short-age that can be relieved by hiring six morepiirchasing clerks. The ABC model, in addi-tion, will trace purchasing costs directly to thenewly designed custom products that are cre-ating the demand for these additional pur-chasing resources, enabling managers to de-termine whether the revenues received fullycompensate the organization for the cost of allthe resources used to produce and deliverthese products.

Of course, supplying additional purchas-ing clerks is only one possible action that themanagers can take to the contemplated activ-ity shortage. The engineering department canbe asked to redesign the custom products sothat they make more use of existing part num-bers, an action that would reduce the amountof additional purchase orders required. Or themanagers can search for process improve-ments or technology that would make the piir-chase order processing activity more efficient,perhaps raising the monthly output per per-son from 125 to 200 purchase orders.

Thus, measuring the costs of resources sup-plied indicates to managers the level of cur-rent spending (or, more generally, expenses)and the capacity to perform activities that thisspending has provided. Measuring the costsof resources used by individual outputs pro-vides information for managerial actions, aswill be discussed more fully subsequently inthe paper.

II. ISN'T THE UNUSED CAPACITYCALCULATION JUST A NEW NAME

FOR THE VOLUME VARIANCE?The calculation of unused capacity each

period looks, at first glance, suspiciously likethe traditional cost accounting volume vari-ance. But the formulas:

Activity Availability = Activity Usage+ Unused Capacity

orCost of ActivitySupplied

= Cost of + Cost ofActivity Unused

Used Activitydiffer from the standai'd cost calculations of avolume variance in several significant ways.

First, and most obviously, volume variancesare reported only in aggregate financial termssince traditional cost systems do not identifythe quantity of overhead resources suppliedor used. The activity-based approach reportsboth the quantity (number of piirchase ordersnot written) and the cost of unused capacity.Second, traditional volxune variances are of-ten calculated with a denominator volumebased on budgeted production, rather thanpractical capacity. In the activity-based ap-proach, the "denominator voliime" must al-ways be the practical capacity of the activitybeing supplied, not the anticipated volume.And, third, the traditional cost accountingprocedure of allocating overhead with a de-nominator volume is viewed as useful only forinventory valuation, not to provide informa-tion relevant for management; e.g..

The preselected production volume level ofthe application base used to set a budgetedfixed-factory-overhead rate for applying coststo inventory is called the denominator volume.In summary, the production volume variancearises because tlie actual production volumelevel achieved usually does not coincide withthe production level used as a denominatorvolume for computing a budgeted applicationrate for inventory costing of fixed-factory over-head.^ (emphasis added)

Note how students are instructed that the cal-culation involves only the application of (so-called) jced-factory overhead to units of pro-duction. Clearly, the volume variance isviewed, at least in textbooks (but not alwaysin practice), as a cost accounting exercise forfinancial statements that is devoid of mana-gerial significance.

These three differences between volumevariances and measurements of imused capac-ity, while real, are not, however, the most im-portant distinction. The cost accounting cal-culation that leads to a volume variance usesa measure of activity volume for the period(i.e., the denominator voliune, also called theallocation base) that varies with the numberof units produced. Direct labor hours, units of

^Charles T. Homgren and George Foster, Cost Account-ing: A Managerial Emphasis, Seventh Edition(Prentice-Hall, 1991), pages 258 and 265.

Accounting Horizons I September 1992

production, materials purchases, and machinehours are typical allocation bases used by tra-ditional systems to assign factory expenses toproducts in production cost centers.^" Implic-itly, this procedure assumes that factory ex-penses are used by products in proportion tothe overhead allocation base, i.e., proportionalto volume of units produced. In practice, ofcourse, this assumption is not valid.

Activity-based cost systems use separateactivity cost drivers (the ABC generalizationof an assignment or allocation base) for eachactivity. The activity cost drivers are not de-vices to allocate costs. They represent the de-mand that outputs make on each activity. Forexample, the activity cost driver for the setupactivity could be the number of setups or thenumber of setup hours; the activity cost driverfor processing purchase orders could be thenumber of purchase orders; the cost driver foradministering and maintaining parts in thesystem could be the number of active partnumbers. While some activity cost drivers areunit-related (such as machine and laborhours), as conventionally assumed, many ac-tivity cost drivers are batch-related, order-re-lated, product-sustaining, and customer-sus-taining.^^

Because traditional cost systems use allo-cation bases that do not represent the de-mands for support resources by activities, thevolume variance for a period can be zero evenwhile substantial shortages or surpluses ofcapadty exist for many individual activities.For example, if actual production includes anunexpectedly high proportion of mature, stan-dard products, produced in large batches, thedemands for many batch and product-sustain-ing activities will be well below the quantityof resources supplied to perform these activi-ties and much unused capacity will exist dur-ing the period. Conversely, if the actual pro-duction volume includes a substantial andunexpectedly high number of new, cvistomizedproducts, that are made in very small batches,the demand for batch and product-sustainingactivities may exceed the quantity supplied.Shortages, delays, and overtime may occur inthe batch and product-sustaining activitieseven though the total quantity of units pro-

duced during the period equaled the budgetedor antidpated amount.

The distinction between the measiirement,by activity-based cost systems, of the cost ofactivities used (and unused) and the tradi-tional cost accounting emphasis on fixed ver-sus variable costs can be reconciled by exam-ining closely the way managers contract forand supply resources to perform organiza-tional activities.

III. RESOURCES THAT ARESUPPLIED AS USED (AND

NEEDED)Some resources are acquired as needed.

For these resources, the cost of resources sup-plied will generally equal the cost of resourcesused. For example, materials are usually or-dered as needed so that materials expenseequals the cost of materials used. And the costof energy supplied to operate production ma-chines £dso equals the cost of using that en-ergy. Tbmporary employees hired on a dailybasis from employment agencies and employ-ees who are paid on a piece-work or overtimebasis are additional examples. The companycontracts with these workers to produce out-put and the workers are paid only when theyare needed to produce output. Capital suppliedby lenders is another example where the sup-ply and the usage cost are identical (equal-ling the interest expense on the amount bor-rowed). ^

In general, when the organization acquiresa resource from outside suppliers, withoutlong-term commitments, the cost of using the

ccHnplex traditional systems that use multipleallocation bases within the same cost center will havemultiple volume variances, but each allocation baseis still unit-level, driven by the volume of output.

^^The hierarchy of factory expenses was introduced inRobin Cooper, "C!ost Classification in Unit-Based andActivity-Based Manufacturing Cost Systems" (Fall1990), pp. 4-13, and discussed further in Robin Coo-per and Robert S. Kaplan, "Profit Priorities from Ac-tivity Based Costing," Harvard Business Review (May-June 1991), pp. 130-137.

i Of course, the commitment fee associated with a lineof credit is a counter-example, because the cost of sup-plying the resource (the right to borrow) is incurredwhether the resource is used or not.

Activity-Based Systems: Measuring ttie Costs of Resource Usage

resource can equal the cost of acquiring (andsupplying) the resource; for example, when theorganization acquires the resource in spotmarkets. Tbe costs of supplying such resoiux^sare apparently what many people have inmind when they refer to 'Variable costs." Suchresources have no unused capacity. Whateveris supplied is used, or, alternatively, whateveris needed is acquired. This causes the costs ofsupplying the resoxirce to be strongly corre-lated with the quantity (and hence the cost)ofthe resource used.

IV. RESOURCES THAT ARESUPPLIED IN ADVANCE OF USAGE

Organizations commit, however, to mak-ing many other resources available whetheror not the resources will be fully used for cur-rent and future activities. This commitmentcan take several forms. The organization canmake a cash expenditure to acquire a resourcethat provides service for several periods intothe future. The most common example occurswhen the company acquires or overhaulsbuildings and equipment. Such a transactionleads to an expense being recognized in eachperiod during the useful life of the resource,with the organization gaining the capacityprovided by the resource during each suchperiod. The expense of supplying the resourcewill be incurred, each period, independent ofhow much ofthe resource is used. ^

As a second example, the organization canenter into an explicit contract to obtain theuse of a resource for several periods in the fu-ture. For example, a company leases buildingsand equipment, or it guarantees ac(»ss to en-ergy or key materials through take-or-pay con-tracts. In this situation, a cash payment willoccur and an expense will be recognized ineach future period. Again, the amount of thecash payment and associated expense are in-dependent of the actual quantity of usage ofthe resource in any period.

The third, and most important, exampleoccurs when an organization enters into im-plicit contracts, particularly with its salariedand hourly employees, to maintain employ-ment levels despite short-term downturns in

activity levels. In this case, the spending (andexpenses) associated with these employeeswill remain constant independent ofthe quan-tity of work performed by the employees.^*

In each of the three contracting mecha-nisms, the organization acquires units of ser-vice capacity before the actual demands forthe service units are realized. Consequently,the expenses of supplying the service capac-ity from these resoiirces are inciirred (or rec-ognized) independent of usage. This indepen-dence in the short-run between the supply (orexpense) of these resoxirces and their \isagehas led this category of expense to be consid-ered "fixed" with respect to current produc-tion volume and mix.

The separation between the acquisition ofresource capacity and its actual usage arisesfrom economies-of-scale in contracting for re-sources. For example, some service vmits comein lumpy amounts (e.g., physical capacity ofmachines, or the services provided by indi-vidual employees). Managers also find it lessexpensive to acquire some resources on a long-term commitment basis rather than to con-tract continually in spot markets to acquireresource capacity as needed. ^ These issues

are using the word "expense" in its traditional ac-counting sense; e.g., an outflow or other using up cfassets or incurrence of liabilities (or a combination ofboth) during a period from delivering or producinggoods, rendering services, or carrying out other activi-ties that constitute an enterprise's (mgoing msgor orcentral operations (W. W. Cooper and Yuji Ijiri, KoKler'sDictionary for Accountants, Sixth Edition (Prentice-Hall: Englewood Chffs, NJ, 1983; pp. 203-204). TV)avoid confusion associated with finandal accountinginventory valuation procedures that shift some periodexpenses forward in time to be matched against fu-ture revenues generated, we will assume, for purposesof this paper and without loss of generality, that unitsproduced always equal units sold. This enables allperiod expenses to be recognized as expenses in theperiod they are incurred.

i*The actual expenses of providing this capability in agiven period can even exceed the cash outlays in thatperiod. This situation arises when cash paymentsmade in much later periods, such as for vacations,pensions and other post-employment benefits, are at-tributed to the supply ofthe resource during the givenperiod.

^*rhis prior commitment can also be made for strategicreasons; see Pankaj Ghemawat, Commitment: TheDynamic of Strategy (Free Press, 1991).

Accounting Horizons /September 1992

bave been discussed at some lengtb by scbol-ars, sucb as Coase, Cbandler, and Williamson.

Tbrough any or all of tbese three contract-ing mechanisms, tbe organization acquires acapability or capacity to perform activities,and an associated expense of providing tbatcapacity. The first step, therefore, in an activ-ity-based analysis is to estimate both the ex-pense of providing the capacity to perform anactivity (the $25,000 monthly expense to pro-cess purchase orders), and the capacity ornumber of units of service activity that canbe practically delivered (the 1,250 purchaseorders per month) by the resoiirces supplied.The expense of providing the activity capac-ity is divided by the number of available ser-vice units to obtain an estimate of the cost ofsupplying a unit of service of the activity (the$20 per piirchase order cost).

V. MEASURING COSTS OFRESOURCES USED IN A PERIOD:THE ROLE FOR ACTIVITY-BASED

COST SYSTEMSThe distinction between resources sup-

plied as needed and resources supplied priorto (but in anticipation of) usage suggests thata relatively simple system can be used for theperiodic measurement of actual expenses (seeExhibit 1). In this system, short-term contri-bution margin is measured as price (or rev-enues) less the cost of resources acquired asneeded: materials, energy, and short-term la-bor (and overtime). By assumption, the re-maining operating expenses represent re-sources that have been acquired prior to ac-tual usage. The costs of these resources shouldbe unaffected by actual activity levels duringthe period. The periodic income statement canreport, for each activity, the costs of resourcesused for outputs and the costs of resourcesunused during the period.

For management purposes, flexible bud-gets and variance analysis become unneces-sary for these expense accounts. A simple com-parison of actual to budgeted expenses, ac-count by account, will suffice to provide feed-back. ^ Basically, the authorized expenseshave been determined either by prior commit-

ments (acquiring plant, property, and equip-ment; signing take-or-pay contracts) or dxir-ing the annual budgeting process. One manu-facturing manager expressed this point quiteforcefully:

Cost variances are useless to me. I don't wantto ever have to look at a cost variance,monthly or weekly. Once you've decided to runa product, you don't have many choices left.Resources are already committed regardlessof how the cost system computes costs amongalternative processes.

Monthly, I do look at the financial reports....I look closely at my fixed expenses and com-pare these to the budgets, especially on dis-cretionary items like travel and maintenance.I also watch headcount. But the financial sys-tems still don't tell me where I am wastingmoney. I expect that if I make operating im-provements, costs should go down, but I don'tworry about the linkage too much. The orga-nizational dynamics make it difficult to linkcause and effect precisely.*'

Managers may be encoiir^ed to modify theiruse of resources in the short-run based on in-formation on unused capacity. For example,when excess setup capacity exists, they cantemporarily decrease batch sizes. Alterna-tively, managers may he expected to adjustdownward the quantity of resources suppliedwhen substantial amounts of unused capac-ity persist for several periods.

Several organizations, however, not under-standing the important distinction betweenmeasuring the costs of resources supplied (andexpensed) and the costs of resources used,have attempted to use their activity-basedsystems to budget monthly expenses. A goodexample of the problems arising from using

i^This distinction between the financial system requiredfor periodic performance measurement (reporting onactual period expenses) and the activity-based systemreporting on the costs of resource usage underlay thearguments in R. S. Kaplan, "One Cost System Isn'tEnough," Harvard Business Review (January-Febru-ary 1988). A good example of a company that sepa-rated its monthly reporting system from the systemused to estimate the cost and profitability of its prod-ucts is provided by the Union Pacific case study de-scribed in the Appendix.

^''Quote taken from Robert S. Kaplan, "Analog Devices:The Half-Life System," HBS # 9-190-061.

Activity-Based Systems: Measuring the Costs of Resource Usage

EXHIBIT 1Example of ABC Income

SALESLess: EXPENSES OF RESOURCES SUPPLIED AS USED

MaterialsEnergyShort-term labor

CONTRIBUTION MARGIN

Less: ACTIVITY EXPENSES: COMMITTED RESOURCES

Permanent direct laborMachine run-timePurchasingReceivini^nventoryProduction runsCustomer administrationEngineering changesParts administration

TOTAL EXPENSES OF COMMITTED RESOURCES

OPERATING PROFIT

Statement

7,600600900

Used

1,4003,200

700450

1,000700800750

9,000

Unused

200

10050

100200

(100)150700

20,000

9,100

10,900

9,700

1,200

an activity-based system for monthly perfor-mance measurement was documented in theHewlett Packard: Queensferry Ttelecommuni-cations Division case.

Hewlett Packard: QTD Case'^QTD had recently installed a new activ-

ity-based cost system. The system accumu-lated expenses at each process and assignedthese expenses to products with a cost driverdefined for each process (e.g., number of axialinsertions). The system was developed prima-rily to provide process cost information toproduct engineers to help them design prod-ucts that would be less expensive to manu-facture. The system, however, was also usedto monitor production performance. The twofunctions soon came into conflict when pro-duction volume dropped due to the postpone-ment of a major contract. The lower produc-tion volume led to large monthly volume vari-ances because operating expenses could notbe reduced proportionately to the decline involume. The controller commented:

In a perfect world, spending would drop tooffset lower production volumes. However, inenvironments like ours, where we retain ouremployees, it is almost impossible for spend-

ing to be cut back when volume drops in aperiod.

Higher cost driver rates were calculated,based on the lower production volimies, so thatthe accoimts would "dear" each period with-out large volume variances. This change, how-ever, negated the primary purpose of thenewly designed system. With unused capac-ity expenses now loaded on to cost driver rates,the system no longer provided product design-ers with acciirate information on the expensesof activities performed to manufacture theirproducts.

Companies like QTD, that attempt to bud-get expenses each month from their activity-based resource usage model, will end up, eachmonth, with a variance representing the un-used capacity for every activity and resoiircefor which usage and availability are not per-fectly correlated. The unused capacity vari-ance signals only that managers did not ad-just the resource availability level to theamount actually required for the volume andmix of outputs produced that period. It is not

Cooper and Kiran Verma, "Hewlett Packard:Queensferry Tfelecommunications Division," HBS Case# 9-191-067.

8 Accounting Horizons/September 1992

helpfiil, however, to predict spending or ex-pense changes.

Once decisions get made on resource avail-ability levels in the organization, typically inthe annual budgeting and authorization pro-cess, the expenses of supplying most resourceswill be determined for the year (unless man-agers deliberately act to eliminate or add tothe resources). For example, the resourcescommitted to the purchase-order processingactivity will be determined anniially as a fimc-tion of the expected number and complexityof purchase orders to be processed. We wouldnot expect, however, the size of the purchas-ing department to fiuctuate weekly or monthlydepending on how many purchase orders getprocessed diiring a week or a month. There-fore, even when usage of a resource drops, theexpense associated with that resource contin-ues at its previous level. The difference be-tween the costs of resources supplied and thecosts of resources used for producing productsequals the cost of unused capacity for theperiod. ® The difference should not be inter-preted as a change in the cost of performingthe activity.

VI. RELEVANCE FORMANAGERIAL DECISIONS: USING

ABC TO INCREASE PROFITSAn improved costing system is a means to

an end. The goal is to increase profits, not toobtain more accurate costs. How do activity-based cost systems help companies improvetheir profitability? We attempt to answer thisquestion through the simple profit equation:

Profits = Revenues - Expenses

Pricing and Product MixSome companies use their ABC informa-

tion to reprice their products, services, or cus-tomers so that the revenues (resoiirces) re-ceived exceed the costs of resources used toproduce products for individual cxistomers. Forexample, prices are lowered to customers or-dering standard products in high volumes, andprices are raised to customers ordering highlycustomized products in low volumes. Pricingstrategies are part of a broader set of actions

taken by managers to improve profits throughchanges in product and customer mix. Forexample, some companies, experiencing de-clining demand for their standard products,proliferated their product line to offer custom-ized, low-volume varieties. This strategy wasinfluenced by their belief that many costs were"fixed" and that the lost volume in standardproducts needed to be replaced with custom-ized products that could "absorb overhead"and even sell at price premiums. With thistraditional view, the labor hours, machinehours, and materials purchases could be ap-proximately the same between the old prod-uct and the new product mix. But the newproduct mix included many customized, lowvolume products that made many more de-mands on resources performing batch andproduct-sustaining activities. Because suffi-cient unused capacity did not exist to performthese activities, the companies had to increasetheir spending so that more resources couldbe supplied to perform batch and product-sus-taining activities. After the product prolifera-tion had occurred, and the companies wereincurring higher expenses for support re-sources, ABC models revealed that many ofthe newly-added products were unprofitable.^"

Once this situation has been discovered,managers have typically first attempted toraise prices on the unprofitable products. Ifthis action does not generate sufficient rev-enues to cover all their product-specific costs,managers contemplate eliminating unprofit-able products. Or they consider outsourcingproducts to suppliers whose total cost of ac-quisition is below the cost of resources re-quired to make the product internally. Ofcourse, before outsourcing or dropping prod-ucts, managers should verify that they caneliminate the resources no longer needed orcan replace the lost volume with more profit-able business. Thus before any decision istaken from activity-based product or customer

i^During a period when usage exceeds ncninal capacity,the difference will represent a "favorable" over-utili-zation of capacity.

^^Unprofitable products are those for which the expensesassigned to maintain, produce, and deliver them ex-ceed the net revenues received from their sale.

Activity-Based Systems: Measuring the Costs of Resource Usage

costs, managers must assess the incrementalrevenue and spending consequences.Critics of ABC have stated:

Isn't this what we have been teaching (orpracticing) as relevant costing or incremen-tal analysis? Students in introductory costand managerial accounting classes are al-ready taui^t that costs unaffected by whethera particular product is retained or eliminatedare irrelevant for that decision and should beexcluded from the analysis. Why do compa-nies need an ABC system? Why not just cal-culate the changes in spending that wouldoccur for any contemplated decision, such asdropping or outsourcing a product, and makea decision based on that analysis? What pur-pose is served by building, maintaining, andattempting to interpret a generalized activ-ity-based cost model?

Perhaps one can understand the demandfor a generalized (activity-based) resource us-age model from a similar situation that arisesin physics. Introductory physics courses teachNewton's laws of motion, such as conservationof angular momentum or gravitational attrac-tion. The principles are illustrated with prob-lems that require calculating the interactionsamong two or three ohjects. Students who sur-vive to more advanced physics courses encoxm-ter a suhject called statistical thermodynam-ics, which provides predictions of the aggre-gate behaviors of large numhers of particles.A naive student might ask, "Why do we needto study thermodynamics as a separate sub-ject? Don't Newton's laws of motion still ap-ply to these particles?" The answer is, ofcourse, they do, but to apply Newton's laws tothe large numbers of particles being studiedwoiild exceed the lifetime and computationalpower of the universe. Therefore, physicistshave devised laws to describe and predict theaggregate behavior of large numbers of inter-acting particles.

"Relevant costing" or "incremental analy-sis" situations are illustrated in introductorycourses and books by simple examples withtwo or three products and simple overheadstructures. An activity-based resource usagemodel can be viewed as the thermod3aiamicequivalent to the three product examples ofintroductory cost accounting courses. Con-sider, for example, the analysis that arises inthe Bridgeton Industries case. ^ The plant ini-

tially produced five product lines. Because ofcompetitive pressures, the plant's profitahil-ity had declined. Special studies were per-formed and eventually two product lines wereoutsourced. As the case proceeds, studentslearn that the total spending on resources de-clined by less than the loss in revenues so thatthe economics of the plant had deterioratedfurther. From a "relevant costing" perspective,how many special analyses would have heenrequired to determine which product lines orcomhinations of product lines should havebeen dropped. Certainly each product line in-dividually could have been analyzed. But be-cause most resources come in lumpy amounts,perhaps suhstantial reductions in resourcesupply (£ind therefore spending) would occuronly if at least two product lines were dropped,as was actually done. But why stop at two?Why not consider dropping all comhinationsof three, or four, or even all five product lines?In total, 2^ or 32 combinations would have tobe analyzed, with the relevant costs calculatedfor each of the 32 possihle maintain/drop com-binations.

The 32 possibilities may not seem insuper-able, but for companies with hundreds andthousands of products, customers, processes,and facilities, the combinations, while still fi-nite, would, as in thermodynamics, exceed thelifetime and computational power of the uni-verse to enumerate much less evaluate. Andretain versus drop is a relatively simple bi-nary decision. What about shifts in productmix, improvements in production processes,and changes in product designs? Managerscannot possibly apply introductory cost ac-counting relevant cost calculations to all pos-sible product and customer mix decisions. Theactivity-based cost model, like the thermody-namics model, provides an aggregate view ofthe economic laws of motion of a complex en-terprise, with thousands of individual prod-ucts, customers, and facilities.^^

^ Robin Cooper, "Bridgeton Industries: Automotive Com-ponent and Fabrication Rant," HBS Case # 9-190-085.

' And even the thermodynamic extension is now knownto be an approximation that ignores relativistic andquantum mechanical phenomena. Similarly, the ac-tivity-based resource usage model, as currently for-mulated, is likely just a first order, linear approxima-tion to what may require stocheustic, nonlinear formu-lations in certain situations.

10 Accounting Horizons / September 1992

Borrowing another analogy, integral cal-culus teaches us that the sum total of doinglots of little things can amount to somethingsubstantial. An activity-based resource usagemodel forecasts the changes in aggregate de-mands for activities from making decisions onmany products, services, and customers. Ineffect, the activity-based cost model performsthe integral calculus function of adding up alot of small effects into something quite sub-stantial. It approximates the changes in re-soiux;e demands that will occur from imple-menting new decisions on pricing, productmix, and customer mix. Before actually imple-menting the proposed decisions, of course,managers must assess the cash flow conse-quences by forecasting, as well, the increasesand decreases in resource supply (includingrevenues) that they anticipate will occur. Anactivity-based cost model serves to direct man-agers' attention to where more detailed analy-sis will likely yield the highest payoffs. TheABC model reduces the dimensionality of de-cisions to where the cash flow consequencesfrom only a few alternatives need to be exam-ined closely.

Change Resource UsageIn addition to pricing, product and cus-

tomer mix changes, which affect profits di-rectly through changes in the margins earnedbetween revenues received and resources ex-pended, ABC models can help managers re-duce resource usage, while holding revenuesconstant. When resource usage is reduced,some unused capacity will be created whichcan then be either managed away (enablinglower spending to occur) or used to processmore throughput (enabling more revenues tobe earned). Demands on support resources canbe reduced by taking two types of actions:

• Reducing the number of times activitiesare performed, and

• Increasing the efficiency with which activi-ties are performed. ^

Reducing number of times activities areperformed:

Changing from unprofitable to profitableproduct and customer mixes, as described

above, enables companies to earn the same oreven higher revenues while performing feweractivities. Managers can take additicmal ac-tions to reduce the number of times activitiesare performed, especially activities performedby support resources. Marketing and salesexecutives in some companies have set mini-mum order sizes to reduce the large numberof activities triggered by many small orders.As engineers improve the design of products,fewer engineering change notices are required.Other change activities are reduced when en-gineering managers discourage their employ-ees from excessive tinkering with existingproduct designs, and marketing managers dis-courage or charge premiums for customer-re-quested changes in products and deliveryschedules. In addition, design engineers, in-formed about the resource expenses associatedwith introducing and maintaining a largenumber of parts in the system, can developproduct designs that use fewer and more com-mon parts.^ All these actions, individuallyand in combination, reduce the number of de-mands for activities performed by support re-sources, while maintaining existing (unit-driven) production volume.

Increasing efficiency (lowering the cost)of activities performed: "

A complementary set of actions can betaken to increase the efficiency of performingactivities. The increased efficiency enables thesame quantity of activities to be performedwith fewer resources. Continuous improve-

^^These actions are iterative, not sequential, as manag-ers continually a^ust the volume and mix of theiroutputs, and manage the efficiency with which theiractivities are performed.

^^These design activities were the focus of the ABC sys-tems described in the Dektronix and Hewlett Packardcases: Robin Cooper and Peter TXimey, "Tfektronix (A),"HBS Case # 9-188-143; and "Hewlett-PackardRoseville Networks Division," HBS Case # 9-189-117.

'^Using activity-based information to focus improvementactivities was discussed in H. Thomas Jolmson, "Ac-tivity-Based Information: A Blueprint for World-ClassManagement Accounting," Management Accounting(June 1988), pp. 23-30. Using an activity-based costsystem for performance improvement was a centra]focus in the system described in Robert S. Kaplan,"Maxwell Appliance Controls," HBS Case # 9-192-058.

Activity-Based Systems: Measuring the Costs of Resource Usage 11

ment programs, such as total quality manage-ment and cycle time reduction (just-in-time),reduce the resources required to inspect prod-ucts, changeover and setup machines, andmove and store materials. Successful imple-mentation of continuous improvement pro-grams produces m^or reductions in the de-mands for resources to perform batch andproduct-sustaining activities.

Introduction of advanced information tech-nology reduces by substantial amounts theexpenses of many batch and product-sustain-ing activities. Computer-Aided-Design andEngineering(CAD/CAE) equipment reducesthe expenses of designing products and mak-ing changes to existing products. They alsostandardize the maintenance of routings andbills-of-materials. Flexible ManufacturingSystems (FMS) and Computer IntegratedManufacturing (CIM) essentially eliminatemany batch activities through automaticscheduling, materials movement, inspection,and tool positioning, gauging, and mainte-nance, plus instantaneous changeovers be-tween operations. In the theoretical limit, aCIM system requires the same resources tomake 1 vinit of 1,000 different products as itdoes to make 1,000 units of 1 product.^* Elec-tronic Data Interchange (EDI) and ElectronicFunds Transfer (EFT) link companies withsuppliers and customers, greatly reducing theexpenses associated with purchasing, sched-uling, receiving, shipping, invoicing, and pay-ing for materials and products.

Improvii^ ProfitsThrough a combination of reducing the

quantity of activities performed and increas-ing the efficiency of performing the remain-ing activities, companies can maintain produc-tion throughput and, hence, revenues whilereducing their demands for indirect and sup-port resources. Ideally, managers can now ob-tain additional business, many of whose de-mands would be handled by resources cur-rently in excess supply. This would enable thecompany to enjoy substantially higher profitsbecause revenues would increase with onlymodest spending increases.^'' Alternatively,

the unused capacity created can be reducedin the next budgeting cycle.

Budgeting: Changing the Supply ofResources to Match Resource Demands

As managers adjust their product and c\is-tomer mixes, introduce new products, phaseout mature products, improve operating pro-cesses, and introduce new technology, theychange the demands for activities performedby indirect and support resources. The reviseddemands for resources to perform support ac-tivities can be estimated with an activity-based model. Differences between the demandfor and the supply of resources can then betranslated into expected changes in futurespending on resources. Used in this way, theactivity-based model becomes a central tool formanagement planning and budgeting. Thebudgets for each resource are determinedbased on the activities required for the fore-casted product volume and mix, and existingproduction processes. For resources forecastedto be in short supply, the analysis provides ajustification for additional spending to in-crease resoiirce availability. For a resourceforecasted to be in excess of predicted de-mands, managers can be requested to reducethe availability and hence the expenses of thatresource. They can reduce the luiused capac-ity by selling or scrapping machinery withoutreplacement, by not replacing employees whoretire or leave the organization vol\intarily, byredeploying employees from activities wherethey are no longer needed to activities wherecapacity shortages exist, or, more drastically,by laying off now redimdant employees. Theseactions enable the company to generate thesame revenues with fewer resoxirces, therebyallowing profits to increase.

effect, CIM transforms batch and product-sustain-ing activities into unit-level activities so that productvariety costs approach zero.

^'Spending will increase for resources for which avail-ability and usage are tightly coupled (e.g., materials,energy), and for resources where unused capacity doesnot exist (perhaps direct labor or machine time). Also,it would be preferable for the added volume to gener-ate revenues in excess of the expenses of resourcesused so that the new business can be sustained in thelong run.

12 Accountir^ Horizons / September 1992

Alternatively, companies may not exploitthe profit opportunities fi*om having createdunused capacity. They may keep existing re-sources in place, even though the demands forthe activities performed by the resources havediminished substantially. In this case, andonly in this case, will the actions that reducedactivity usage not yield any tangihle benefits.Profits will remain the same, since revenueshave remained constant and the expenses ofresources supplied have also remained fixed.But the failxire to increase profits is not dueto costs being intrinsically "fixed." Rather, thefailure is the consequence of managers beingunable or unwilling to exploit the unused ca-pacity they have created. The activity-basedcost model focuses managers' attention ondecisions that affect the resource demands byactivities. If tbe decisions lead to lower de-mands for some resources, the company canthen realize increased profits by either usingthese resources to generate higher revenuesor by reducing spending on these resources.The costs of these resources are only "fixed" ifmanagers cannot or do not exploit the oppor-tunities firom the unused capacity they helpedto create.

VII. SUMMARY ANDCONCLUSIONS

Activity-hased cost systems contain twoimportant insights. First, the activities per-formed by many resources are not demandedin proportion to the total volume of units pro-duced (or sold). The demands arise from thediversity and complexity of the product andcustomer mix.

Second, activity-based cost systems are notmodels of how expenses or spending vary inthe short-run. ABC systems estimate the costsof resources used to perform activities for vari-ous outputs. During any given period, the pro-duction of products and services, and theirmarketing, sale, and delivery to customers,create a demand for organizational activities.The quantity of each activity supplied to out-puts is estimated by activity cost drivers suchas the niunber of setup hours, nxunber of pur-chase orders processed, number of receipts,number of direct labor and machine hours, andnumber of parts maintained. By summingacross the costs of all resources supplied toperform activities for individual outputs, theABC model estimates the costs of resourcesused during the period by all the organi-zation's outputs.

Activity-based systems model how activ-ity usage varies with the demands made forthese activities. If activity usage exceeds thequantity available from existing resource sup-ply, then higher spending to incres^e the sup-ply of resources will likely soon occur. If, how-ever, activity usage is below available supply,spending or the expenses of resources will notdecrease automatically. Management, to ob-tain higher profits, must take conscious ac-tions either to use the available capacity tosupport a higher volume of business (i.e., byincreasing revenues) or to reduce spending onresoiirces by eliminating the unused capacity.Costs and profits are fixed only if managementtakes no action, and leaves the unused capac-ity undisturbed. Management behavior, notcost behavior, determines whether reductionsin resource demands become translated intobigher profits.

Activity-Based Systems: Measuring the Costs of Resource Usage 13

APPENDIXSeparate Systems for Measuring Resource Expenses and Resource Usage:

A Case Study

The Union Pacific case study illustrates well how a service organization developed a system formeasuring the costs of resource usage quite different from the system used for operational and expensecontrol.^ During the 1960s, the company had developed an extensive system for monitoring spendingand expenses in its more than 5,000 cost centers around the country. Cost centers included freight andlocomotive repair yards, switching yards, transportation crews, and maintenance of track and rig^t ofway. Expenses were recorded in up to 1,200 different account codes.^ Each month, a cost center managerreceived a report on actual and budgeted expenses for each of these accounts, supplemented with data onYear-tO'Date actual expenses compared with budget and with a similar period in the previous year. The5,000 individual cost center expense control reports were agsgregated into summary data for higher levelmanagers all the way to senior vice-presidents in Omaha who received a one page summary of operationsunder their control. This extensive system of monthly reports was used to monitor and control cost centerexpenses and measure efficiency improvements.

In the deregulated environment of the 1980s, the company realized that despite extensive reportingof cost center expenses, it had no information to estimate the costs of resources used to move a carload offreight from one point to another. This gap occurred for two reasons. The railroad environment provides avivid example of where almost complete separation exists between resource spending and resource usage.The monthly spending to maintain track and right of way and to repair locomotives and freight cars hasno relation to the amount of traffic run that month. The monthly spending reflects the millions of grosston miles hauled in many preceding months, and management's dedsion to replenish the supply of theseresources so that they will be available for the future. The cost of using those resources occurred in thepast; the spending to revitalize the depleted resources was occurring today.

Even apart from the temporal separation between resource usage and resource spending, the railroadlike many other service organizations did not measure the use of resources by individual products withineach cost center. For example, the railroad supplied switching yards and measured the expenses of operatingswitching yards. But it did not measure the quantity of use of switching yards by individual freight carsas they moved from shipper to customer.

The railroad had to develop entirely new analytic systems to measure the costs of activities performedto supply its customers with products and services. The costs of resources used to move a carload offreight from shipper to destination could not be estimated based on incremental spending since virtuallyno incremental spending occurred when the company picked up a freight car from a shipper, scheduled it,connected it to a train, switched it to several different trains, and finally delivered it to the customer. Yetthe movement of the freight car required an extensive quantity of railroad resources to be supplied andavailable. And the actual running of the freight car placed incremental demands on several resourcesthat would require additional spending sometime in the future. The comptuiy understood that it could notwait until the freight car, locomotive, or track was repaired to send out bills to all the shippers that madeuse of these resources in the past. It also understood that the amounts spent to supply train crews,scheduling and information systems, and switching yards were justified by the expected volume and mixof trjiffic to be carried. The company developed a system that estimated, move by move, the quantity andcost of all the resources used by individual carload moves, even though short-run spending was almostcompletely independent of these moves.

The railroad example provides a vivid example of the difference between resource usage and resourcespending (or resource expenses). The power of the case, however, extends beyond railroads or even servicecompanies since most manufacturing companies' resources are also now characterized by large distinctionsbetween the use of the resources and the amount of current expenses to supply the resources.

. Kaplan, "Union Pacific: Introduction, (A), and (B)," HBS Cases 9-186-176, -177, -178.^*rhe larger number of account codes arose fi-om regulatory reporting requirements specified by the Interstate Com-

merce Commission for Railform A.