A fast way to decide if your cost systemgives you bad information

You Need aNew Cost System Whenby Robin Cooper

By now it's well publicized-if not obvious-thatmany companies' cost accounting systems are fallingdown on the job. They give managers incorrect prod-uct costing information, or they inundate managerswith irrelevant cost information, or they fail to mea-sure the things that really count. Strategies may heconceptually brilliant, but if they are based on faultyinformation about the cost of a product, they arelikely to fail in the marketplace. Many have.

But redesigning a cost system is expensive andtime consuming. Do you really have to do it? Thereare two ways of finding out.

An obsolete cost system sends many signals, soone way to discover if you need a new system is toleam how to read those signals. [See the insert for adefinition of an ohsolete eost system.)

Cost systems don't heeome ohsolete overnight.They gradually outlive their usefulness as they fail toadapt to change. So a second way to tell if your sys-tem has deteriorated is to analyze the changes thathave occurred in your organization and in its envi-ronment since you first implemented the system.

It's Time to Redesign YourSystem If YouNotice That...

...functional managers want to drop seeminglyprofitable lines. Production managers know when aproduct is troublesome. And marketing managers

know when a product isn't priced competitively. Youcan use their knowledge to test your cost system.Ask them to list the ten established products theywould most like to drop. If there is nothing specialahout those items, and yet they still show high prof-its, the cost system may he failing to capture theirtrue complexity.

...profit margins are hard to explain. Managersshould he able to give simple explanations of profitmargins: "We have the best production technolo-gies"; "We have lower production volumes"; "No-hody else makes that product"; or "We set thestandard and make a premium for doing so." In onecompany, the production manager was under con-stant pressure to make a certain new product morecheaply. He couldn't explain the high costs. He wasconfident that he was doing a good joh and believedthe product should he competitive. Years later a re-vised cost system showed that because the productused more direct labor than any other, it was beingcharged too much overhead. It was in fact the com-pany's most profitable line. Unfortunately, by then,competitors had introduced similar products and theopportunity was lost.

...hard-to-make products shov/big profits. A goodtest of a cost system is an item that's harder to makeor requires more inspection or rework than others.

Robin Cooper is associate professor of business adminis-tration at the Harvard Business School and a fellow ofthe Institute of Chartered Accountants in England andWales.

HARVARD BUSINESS REVIEW lanuary-February 19X9 77

What Is an Obsolete Cost System?A cost system sbouldn't necessarily measure abso-

lutely everything down to the finest degree. Taking in-finitesimal measurements of eacb bit of material andeach second nf direct labor ean be expensive and timeconsummg. The expense is necessary only when theconsequences of relying on inaccurate information aresevere. When, for instance, margins are paper thin and

The optimal cost systemminimizes total costs and

= Optimal cast systemAccurocy •

... changes when informationtechnology improves...

Accuracy >• = Current values = Prior values

tbe market moves quickly, basing decisions on inac-curate cost data can put a company out of business in ahurry. In other situations, highly accurate numbers areless important, and tbe company sbouldn't spend a lotof money to get them.

A good cost system trades off the cost of measure-ment and the cost of errors from inaccurate informa-tion in a way that minimizes total cost (see accom-panying graphs). As an economist would put it, theoptimal system exists at the point where the margi-nal cost of improving the system's accuracy exactlyequals the marginal benefit.

An optimal cost system is a moving target. Competi-tive conditions are dynamic, so the cost of errorschanges. Similarly, as information-processing technol-ogy changes, so does the cost of measurement.

It is important to rememher that product diversityhas a great deal to do with accuracy. As diversity in-creases-as high volume is mixed with low volume, oriahor intensity is mixed with automation-costs arcmore likely to he skewed. To achieve the same level ofaccuracy, companies will have to spend more on mea-surements than when products were more homogene-ous. If they don't, their cost systems will he ohsolete.

... or when errors becomemore costly.

Accuracy

Sueh products will have higher than average costsand, unless they are priced at a premium, will havelow margins. If they are not premium priced hut ap-pear to be highly profitahlc, the cost system is failingto report their true cost.

...departments have tbeir own cost systems.When functional managers have completely lostfaith in the official cost system, they may develop

systems of their own. Personal computers make itfairly easy to do. The design engineers in an electron-ics company didn't trust the numhers the cost sys-tem produced. Bad or complex designs came outlooking like hig profit makers, while products the en-gineers knew to be well designed appeared to be los-ers. The engineering department responded hydeveloping its own system for costing products.

78 HARVARD BUSINESS REVIEW lamiary-February 1989

COST SYSTEM

Where the official system used direct labor to allo-cate costs, the private system used a number of differ-ent bases. Also, the engineers tracked costs theyconsidered to be product related but that the officialsystem treated as period expenses. The departmentignored the official system and used its private sys-tem to steer design work.

,..the accounting department spends a lot of timeon special projects. Some decisions require more ac-curate information than others. A decision on out-sourcing a bigh-volume product, for example, isimportant enough to warrant more detailed and ac-curate cost data. Accounting departments often setup special teams to study such situations. But thecost system-if it's doing its joh-should provide

Managers can quicklydiagncse an cbsclete castsystem by checking thesymptoms.

managers with much of the information they need. Ifits failure to do so makes lengthy special studies rou-tine, the cost system is probably obsolete. This wasthe case at one company, where half of the account-ing staff was working on special projects, some ofwhich took more than six months to complete.

...you have a high-margin niche all to yourself.Unless barriers to entry exist, companies should ex-pect competition. If there is none, the cost systemmay be reporting fictitious margins. One companyfound that as its niche expanded, overall margins fell.A redesigned cost system showed that products thecompany thought were earning high profits were ac-tually losing money. Another company discoveredthat a competitor was buying its products and tbenrepackaging and selling them. The company's sellingprice, based on faulty cost information, was lowerthan the competitor's production cost.

,..competitors' prices are unrealistically low.When other successful companies, especially smallerones, charge less for items you produce in high vol-ume, your cost system is suspect. It is likely the sys-tem averages product costs among your high- andlow-volume items. The smaller company probablymakes products whovSe production volumes are simi-lar, so averaging creates less cost distortion.

,..customers don't mind price increases. Custom-ers will certainly never ask for price increases. But ifthey aren't surprised wben increases come, they mayknow more about your costs than you. Even if they

1, Sec Robert S, Kaplan, "One Cnst System Isn't Enough," HBK lanuary-Fcbruary 1988.p,61,

complain, they may think, "It's about time" and paythe higher prices willingly. After all, they may haveexplored making the product themselves or have in-formation on competitors' costs. When one companysensed that its prices were too low, it raised them by25%. The market didn't flinch. Customers paid thehigher prices without complaint; sales volumedropped off only sligbtly. The market confirmedmanagement's intuition that the cost inft)rmationwas wrong.

,.. the results of bids are hard to explain. Unless themarket is chaotic, managers should be able to esti-mate the competitiveness of their bids. In particular,they should be able to set high bids for business theydon't really want and low bids for business that's im-portant to them. But a company's bids are often basedon the cost information it gets internally. If that in-formation is faulty, the company will have no ideahow its bid compares with competitors; When onecutting-tool manufacturer kept winning high bidsand losing low bids, the president suspected that thecost system was the culprit. A new cost systemsolved the problem.

...vendor bids are lower than expected. Compa-nies that are considering outsourcing can comparevendors' bids with their own costs to tell if their costsystem is working well. If the bid price varies widelyfrom the cost of making the product, the cost systemmay he at fault. In one ease, the vendor bid was belowthe variable cost of the product, yet there were no in-dications that the vendor was that much more effi-cient. A special team looked into the matter andfound that the product could in fact be made morecheaply in-house-despite the cost system's messageto the contrary.

...reported costs change because of new financialaccounting regulations. Systems designed with onegoal in mind generally don't do a good job of meetingothers.' A system that aims to meet financial report-ing requirements probably distorts costs. If a newGAAP regulation changes your costs even when ma-terials prices and manufacturing costs are constant,chances are you tailored your system to meet finan-cial reporting requirements-not to provide accuratecost information.

Your System May Be Obsolete It You'veExperienced...

...increased automation. When direct labor is usedas an allocation base, the introduction of automatedproduction processes such as flexible machining sys-tems can cause the system to fail. The new machin-

HARVARD BUSINESS REVIEW January-Febmary 19S'; 79

COST SYSTEM

ery uses less direct labor but usually requires moresupport for programming and engineering. Productsmade through automation tend not to be chargedenough overhead, while products manufactured con-ventionally are charged too much.

One company had completely revamped its pro-duction process to move from machines that re-quired continuous direct-labor supervision to ma-chines that required virtually no operator atten-dance. The direct-labor-based system failed to cap-ture the economics of the new production process be-cause it didn't allocate overhead to products made onthe new machines, A system based on machine hourscorrected tbe problem. Reported costs of some prod-ucts changed hy as much as 30%.

IIt's time for a new cast systemwhen engineeringdevelops cne af its awn.

An integrated-circuit manufacturer introduced anew line of chips, which was a good strategic fit butdid not fit well with the cost system. Production ofthe new product was highly automated, so when itcame time to allocate overhead, as always, on the ba-sis of direct labor hours, the new product got off easy.The system didn't reflect the new line's more inten-sive use of the costly automated machines. The costsystem was distorted. Even worse, the existing prod-ucts were charged witb too much overhead and ap-peared unprofitable. Subsequently, the companymoved their production offshore.

...changes in the use of support functions. If anewproduct requires different kinds of support from ex-isting lines-more detailed inspection, for example,or longer setups-the amount of overhead allocatedto it will likely be incorrect. These distortions cancreep in slowly. For example, one company intro-duced a new line of plastic products to complementits sheet metal business. Initially, volume was rela-tively low, so little distortion arose from allocatingthe overhead needed only for metal fabrication toboth plastic and metal products. Over time, however,sales of the plastic products increased dramatically,and the distortion became serious.

...changes in product market strategy. The deci-sion to market in a low-volume niche means smallerproduction volume. In contrast, the decision to movefrom experimental parts to production parts meanshigher volume. Most cost systems are designed withone type of production in mind and don't differenti-ate well between tbe overhead consumed by high-and low-volume products. When production vol-

umes vary widely in the same company, cost distor-tion arises. If production volumes are fairly similar-say, volume of one product is no more than fivetimes that of any other-product costs will probablybe accurate. Accuracy falls off rapidly as the rangegrows to more than 10 to 1.-

One company produced some products in batchesof under SO and others in batches of more than 1,000.Its traditional direct-labor-based cost system grosslyundercosted the low-volume products and madethem appear more profitable than they were. Thecompany thought its product strategy-trying to heeverything to everybody - was working, but the eco-nomies were misleading. Year after year, profits fell,and the company was eventually taken over.

In contrast, another company was forced to adoptthe strategy of producing low-volume, customizedproducts because competitors had an overwhelminglabor cost advantage. To ensure that the orders it ac-cepted for low-volume items were truly profitable,the company introduced a new cost system thatmore appropriately traced overhead to high- and low-volume products. The new cost system helped thecompany implement its new strategy successfully.

...simplification of manufacturing processes.Changes in the production environment don't neces-sarily require more complex cost systems. The intro-duction of new and simpler production philosophies,such as just-in-time (JIT) or cellular manufacturing,can make a needlessly complex system obsolete. Inone company, the cost system measured the valueof work in process at every inventory stage, requir-ing hundreds of thousands of measurements a year.But the introduction of JIT reduced inventory lev-els so much that those measurements were nolonger important.

Cellular manufacturing has the same effect on oldcost systems. This manufacturing approaeh creates aseries of mini-factories, each specializing in similaritems. Companies should be able to trace overheaddirectly to the mini-factories and then spread thosecosts evenly over all the units they produce, A costsystem that traces costs to individual products isprohably obsolete.

...intensified competition. When competitionheats up, so does the chance that a competitor willtake advantage of a poor pricing decision. The in-creased risk associated with poor cost informationcan make a system obsolete. When a product is over-costed, its profit margin will look deceptively unat-tractive. If a competitor gives chase to the produet,the company may mistakenly decide not to defendits position. Alternatively, prices set too high because

I. Sec Robin Cooper and Robert S, Kaplan, "Measure Costs Right: Makethe Right Decisions," HBR SL-ptfmbcr-Octobtr 19^8, p, 96,

HARVARD BUSINESS REVIEW lanuary February 1989

they're overcosted might attract competitors thatwould otherwise have faced a higher barrier to entry.One company redesigned its cost system and discov-ered that a particular product line was considerablymore profitable than it had thought. To avoid attract-ing competition, the company increased the dis-count, added more field support, and increasedadvertising spending.

When companies eam a reasonable overall margin,they often don't worry about the margins individualproducts make. In the face of stiffer competition,though, management needs reliable cost informationto act confidently Executives must know how muchleeway they have in underprieing the competitionand at what point a product line is not worth saving.

One manufacturing company introduced a newsystem just in the nick of time. Its old system hadbeen generating huge year-end variances, so the pres-ident astutely targeted the eost system for redesign.Soon after, the industry went into a slump and pricesfell dramatically The company, with its new knowl-edge of product costs, was ahle to cut prices more

FIRST AMERlCflN BflNKOFTHE SOVIET UNION

aggressively than other players. It picked up thebusiness of several failed competitors and in certainproduct lines increased its market share perma-nently hy as much as 300%.

...unbundling of products. For many years onecompany had bundled two products together: ap-parel fastener machines, which the customer rented,and apparel fasteners, which the customer attachedto their products using the machines. The rental feefor the machines was purposely set low to attractcustomers, who became captive buyers of the fasten-ers. The price of the fasteners was set high enough tocover their costs as well as the unrecovered costs ofthe machines. The cost system traced all overheadcosts to fasteners and none to the machines.

On the surface, the system worked fine. Custom-ers were happy and loyal. But when the company re-designed the cost system so it separated costs of thetwo types of products, it became clear that the costsystem was actually sending highly distorted signals.Because some of the fasteners were labor intensive,the old system had attributed a disproportionate

amount of overhead (including thecost of the attaching machines) tothem. Over the years, the companyhad put little effort into these productlines and consequently walked awayfrom attractive markets.

...deregulation. Under regulation, acompany doesn't set the prices; theregulators do. Companies make profitsby controlling overall efficiency. De-regulation increases a company's com-petitive choices hut can make a costsystem useless. When companies havenew freedom to "cherry pick" prod-ucts, accurate knowledge of costs isinvaluable. One railroad company, forexample, when faced with deregula-tion, introduced a new cost systemthat for the first time in the company'shistory reported the cost of a freight-car move from city to city. Its existingcost system reported tbe cost of eachfunction (switching, locomotive re-pair, and the like) but not the cost ofa move. Knowledge of these eosts al-lowed the railroad to compete moreeffectively with other railroads andtrucking companies.

Some situations mimic the effectsof deregulation, like when a captivesupplier is allowed or forced to com-pete on the outside. An internal trans-fer pricing system, for instance, acts

HARVARD BUSINESS REVIEW fanuary-Fcbruary 1989 SI

COST SYSTEM

much like a regulated pricing system. One companythat recently began to compete in the open marketdiscovered that because of a faulty cost system, itspricing attracted the business it was least interestedin and turned away the business it really wanted. Thecompany secured sales of low-volume, complexproducts instead of the high-volume, simple prod-ucts for which its production facility was designed. Anew cost system corrected the prohlem and allowedthe company to bid more aggressively on high-volume business.

...technological improvements. Systems can be-come obsolete if they fail to take advantage of techni-cal improvements that permit more efficient datagathering and analysis. The introduction of a com-puterized, production floor scheduling system, forexample, captures considerably more informationabout the products. This information can go into thecost system at virtually no cost.

Similarly, numerically controlled equipment, es-pecially when controlled by a central computer, in-creases the availability of machine-readable infor-mation. Setup time and run times can he measureddirectly at no extra cost. Remote sensing technol-ogy, such as bar coding, can also provide lots of newinformation at little additional cost.

.,. changes in strategy and behavioral goals. Some-times management changes its strategy and there-fore wants to encourage and reward differentbehavior. The cost system doesn't always adapt. One

IThe system is outdated whenevery decisicn requires aspecial accounting team.

company was successful because of its technologicalinnovation. When the market sent signals that cost-not just technical superiority-was important, thecompany decided to pay more attention to efficientdesigns. It urged its engineers to stop designing fromthe ground up and to incorporate some of the partsalready in use. The old cost system couldn't tracksuch things as how many part numbers were used, sothere was no way to identify expensive productsmade witb low-volume or unique components. Thecompany designed its new system with this impor-tant new variahle in mind.

Another company's strategy moved toward prod-ucts with very short life cycles. Designing productsthat could he manufactured economically in smallbatches became critical. In particular, the company

needed to compare the cost of inserting differenttypes of components both manually and automati-cally. A new cost system enabled it to do that.

Another company bought an expensive piece oftest equipment to improve product quality. But theold cost system treated tbe machine as overhead. Thework force took advantage of this "free" work centerby building complex products on it, A new cost sys-tem ensured that workers put the new equipment tobest use by making the test area a cost center andcharging products an hourly rate for using it.

Is It Time?

The mere presence of symptoms doesn't mean thecost system is obsolete. A product may have inexpli-cably low profit margins because the cost system isobsolete- or because a competitor has adopted a pen-etration strategy. It helps to think about the intemaland external changes that make a cost system obso-lete. They provide more clues to whether your sys-tem needs fixing. Checking for symptoms andlooking for changes that may have caused them givesa good indication of the effectiveness of your currentsystem. If you find no symptoms, the system is doingfine. If you observe several symptoms and knowwhat probably caused them, it's time for a redesign.

The hard part is when you detect only one or twosymptoms. Then the call is more difficult to make.One way to proceed is to set up a pilot cost system fora single product line and compare the numbers withthose the existing system produces. If the results dif-fer widely, a redesign is in order.

Remember that because conditions keep chang-ing, managers should evaluate their systems everyfew years. They don't necessarily have to design anew system that often. Before a company plungesinto redesign it should be sure to analyze the invest-ment. The potential savings—the difference hetweenthe total costs of the existing system and the totalcosts of a new one-should exceed the cost of devel-oping and implementing the new one.

A cost system, with modifications along the way,should last about a decade. But at some point, you canno longer patch up and add on to the system. Compa-nies may not want to face up to the fact that theircost systems need to be redesigned, but if they don't,they may face far more severe consequences. A busi-ness that doesn't know what its products really costwon't be in business for long. ^

Reprint 89102

82 HARVARD BUSINESS REVIEW fan uary-February 1989