One Cost System Isn't Enough: Harvard Business Review
One Cost System Isn't Enough: Harvard Business Review
One Cost System Isn't Enough: Harvard Business Review
os
One Cost System
Isn’t Enough
rP
by Robert S. Kaplan
yo
op
tC
No
Do
Reprint 88106
This document is authorized for educator review use only by Lenin Faican, Universidad Internacional SEK until Aug 2025. Copying or posting is an infringement of copyright.
[email protected] or 617.783.7860
t
HBR
os
J A N U A RY– F E B R U A RY 1 9 8 8
rP
One Cost System Isn’t Enough
yo
Robert S. Kaplan
M
any companies now recognize that their business had an excellent system that promoted cost
op
cost systems are inadequate for today’s control and production efficiency. It yielded frequent
powerful competition. Systems designed reports on direct labor use and efficiency, scrap
mainly to value inventory for financial and tax state- buildup, and department expenditures. The only in-
ments are not giving managers the accurate and formation on product costs, however, came from the
timely information they need to promote operating standard cost system used to allocate overhead for
efficiencies and measure product costs. financial reporting purposes. This system had re-
tC
In response, they have tried to redesign their pre- cently been redesigned so that overhead costs were
sent systems, but results have been disappointing. allocated to products using machine hours and mate-
M One chemical company’s system did a good job rial dollars as well as the traditional direct-labor hour
of estimating full product costs but could not be used base. But even with this redesigned system, the divi-
for cost control. It gathered product costs at each sion’s attempts to seek outside customers were un-
production stage and cumulatively absorbed all vari- dermined by highly distorted product cost estimates.
ances along the production trail. While the system
reported actual costs for all products, it provided no Robert S. Kaplan is Arthur Lowes Dickinson Professor of
No
information to motivate or evaluate the cost-reduc- Accounting at the Harvard Business School and professor
tion efforts of production managers. of industrial administration at Carnegie-Mellon Univer-
As competition shifted to low-cost production of sity. He won the 1987 award given jointly by the American
commodity products, the company had to develop a Accounting Association and the American Institute of
new cost system to give unit managers more reliable Certified Public Accountants for the most notable contri-
information about their production efficiency. Head- bution to accounting literature, ‘’Measuring Manufactur-
quarters scrapped the old system and installed one ing Performance: A New Challenge for Management Ac-
that isolated all variances at the cost centers where counting Research," Accounting Review (October 1983).
His latest book is Relevance Lost: The Rise and Fall of
they occurred. Local managers could now observe the
Management Accounting, with H. Thomas Johnson (Har-
Do
Copyright © 1988 by the President and Fellows of Harvard College. All rights reserved.
This document is authorized for educator review use only by Lenin Faican, Universidad Internacional SEK until Aug 2025. Copying or posting is an infringement of copyright.
[email protected] or 617.783.7860
Why are so many companies having such diffi- tems divide these costs—labor, materials purchases,
culty? Cost system designers have failed to recognize and factory overhead—between items sold and those
t
that their systems need to address three different still in stock. Financial accounting principles do not
os
functions: require that assigned overhead costs be causally re-
lated to the demands of individual products, so many
Inventory valuation for financial and tax state-
companies continue to use direct labor to allocate
ments, allocating periodic production costs be-
overhead, even though direct labor may account for
tween goods sold and goods in stock.
less than 5% of total manufacturing costs. Moreover,
Operational control, providing feedback to
businesses can use a single plantwide burden rate for
production and department managers on the re-
rP
allocating overhead to products, regardless of the
sources consumed (labor, materials, energy, over-
diversity of their production processes.
head) during an operating period.
Therefore, a company’s overhead allocation scheme
Individual product cost measurement.
may not correspond to the underlying production
Even if cost system designers recognize how impor- process or to the demands individual products make
tant and how different the demands of these three on the enterprise’s resources. Auditors won’t ques-
functions are, their efforts are blocked by senior ex- tion cost-of-sales or inventory valuation estimates
ecutives’ insistence on a single “official” system. merely because the company has used an aggregated,
yo
And when compromises have to be made, the de- simplistic method for assigning overhead costs to
mands of the financial reporting function (inventory products. As long as the split of costs between goods
valuation) invariably triumph. The more manageri- sold and goods still in stock is fairly accurate, in
ally relevant functions of operational control and aggregate, the needs of financial reports will have
product costing usually suffer. been met.
Many businesses know the consequences of this The cost system for external reporting does not,
dilemma all too well. Operating costs are reported too however, give managers relevant performance meas-
late and are too aggregated to benefit production urement and product cost information.
op
supervisors. Managers must use product cost esti-
mates that focus on the least important cost compo-
nent—direct labor—and ignore expenses involved in OPERATIONAL CONTROL
designing, marketing, distributing, and servicing
goods. An operational control system must provide accu-
Businesses can no longer afford cost systems that rate, timely feedback to managers on their perfor-
tC
work well only to value inventory for financial re- mance. The system must correspond to the unit
porting. No single system can adequately cover all manager’s level of responsibility, control for known
three functions. The demands of each differ in terms variations in cost behavior, and minimize the inci-
of reporting frequency, degree of allocation, nature of dence of cost allocations. Cost-accounting calcula-
cost variability, system scope, and degree of objectiv- tions (like the allocation of overhead to products and
ity (see the Exhibit). departments, or the computation of volume vari-
The chemical company in the first example was ances) should not be part of a company’s operational
better off than most: at least it had one system sepa-
No
three cost system functions. of parts per hour, the per-unit materials, labor, ma-
chine time, and utility consumption should be re-
ported daily or even hourly. The system for control
INVENTORY VALUATION in a support department or research lab could report
on a much longer cycle.
Under generally accepted accounting principles, Obviously, it is not much help to get monthly cost
manufacturers must allocate periodic production reports for an operation that turns out many parts per
costs to all items produced. Inventory valuation sys- second. A manager controlling work hourly and daily
t
Different Functions, Different Demands
os
Degree of Scope of Nature of Degree of
Functions Frequency Allocation System Variability Objectivity
rP
Operational Daily, by unit None Responsibility center Short-term High
control of work variable and
accomplished fixed
Product cost Annually and at Extensive, down to Entire organization, All variable Low
measurement major change individual products including production,
points or product lines marketing and
distribution,
yo
engineering, service,
and administration
does not want to receive an aggregate variance report Cost Allocations. Many companies routinely allo-
in the middle of the subsequent month. Equally as cate costs to a cost center, even when the center has
obvious, daily or weekly cost reports would confuse little or no control over them. This practice evolved
op
departments taking several months to assemble a because, to value inventory, all factory costs must be
complex machine or performing basic research. allocated to products. With traditional inventory
For operations under computer control, the digital costing systems, plant and overhead costs are first
data can be captured to record what, when, and how allocated to cost centers and then, using a cost center
much was produced. Companies no longer need to burden rate, allocated to products.
collect production data with stopwatches, time Once a company separates its system for measur-
tC
clocks, and clipboards. Automatic bar-code reading of ing operating performance from that used to value
parts combined with local area networks permit con- inventory, however, it does not have to allocate com-
tinual tracking of parts and operations. Cost control mon or noncontrollable costs to individual cost cen-
systems can record these data and provide frequent, ters. Only those costs that are directly related to
accurate reports on actual output and resource con- actions taken within a cost center and whose con-
sumption. sumption can be accurately measured at the cost
center level should be reported periodically to the
Cost Fluctuations. Effective operational control unit manager.
No
requires an understanding of which costs are fixed For example, a cost center’s metered demand for
and which change with short-term variations in ac- kilowatt-hours of electricity or pounds of steam
tivity. Separating costs in this way permits prepara- should be assigned to that center. But if metering is
tion of flexible budgets that adjust for changes in difficult, a company does not improve cost control
activity levels on the consumption of labor, materi- activities by allocating a factorywide utility expense
als, machine time, energy, and support services. to cost centers.
It is easier to establish a flexible budget for opera- By avoiding allocations, the operating report can be
tional control when analysts grasp the underlying based on accurate, objective data on the cost center’s
scientific or engineering laws governing the produc- consumption of resources during a period. Ballpark
Do
tion process. They can then build the cost control estimates of the quantity of labor, machine time, and
system on the production standards established by support resources used don’t help managers’ effi-
the conversion process. A production process that is ciency and productivity improvement efforts. More-
stable and repetitive also helps to predict the relation- over, operating reports filled with estimated and allo-
ship between inputs and outputs. In both cases, the cated costs distract cost center managers from their
company can base its cost control system on a flexible primary responsibilities to monitor and control pro-
budget that adjusts for costs that vary with fluctua- duction efficiencies and to improve productivity. If
tions in short-run production activity. headquarters occasionally needs unit managers to
t
cost center—on a one-time basis, for information costs usually bear no relation to the resources con-
os
purposes only. sumed to design, produce, market, and deliver the
product. I have seen cases where a more accurate
Nonfinancial Measurements. Cost information system revealed that products yielding healthy prof-
may, in fact, play only a minor role in operational its according to the standard cost system—with indi-
control. A company maintains control best at the cated margins of more than 45%—were actually los-
shop-floor level by frequent reports of measures like ing money. Similarly, careful analyses of marketing
yield, defects, output, setup and throughput times,
rP
and distribution expenses have shown that product
and physical inventory levels. At the department lines, previously considered to be only breaking even,
level, monthly summaries of quality control (part- were actually among the company’s most profitable.
per-million defect rates, percentage of items pro- Seriously distorted product costs can lead manag-
duced with no rework required), average throughput ers to choose a losing competitive strategy by de-em-
times, percentage of delivery commitments met, in- phasizing and overpricing products that are highly
ventory levels, new product introduction times, and profitable and by expanding commitments to com-
marketing and distribution statistics make up the plex, unprofitable lines. The company persists in the
yo
most relevant set of performance measures. Financial losing strategy because executives have no alterna-
measurements are useful for periodically comparing tive sources of information to signal when product
actual with budgeted expenditures in each depart- costs are distorted. Only after many years of declining
ment. Measures of process costs will be helpful when market share and reduced profitability will managers
many inputs are combined into intermediate and learn how erroneous product costs led to poor product
finished products. But many companies rely too mix and pricing decisions.
much on summary financial measures and ignore the Analysts, attempting to understand the demands a
powerful opportunities for continual improvement product makes on the company’s resources, can start
op
that a well-constructed set of nonfinancial operating by interviewing the supervisors of production, sup-
measures can give them. port, logistics, and marketing departments. They
must learn what creates work for the resources in
these areas, the cost of performing the work, and the
PRODUCT COST MEASUREMENT quantity of work demanded by individual products.
tC
Even the best designed and implemented operational Allocations and Estimates. Extensive allocations
control system, however, can be useless for measur- of support department costs may be necessary to
ing product costs. Take the experience of one com- estimate the unit costs of the activities that these
pany in the transportation industry. By the late 1960s, departments perform. In the transportation company,
the company had developed an extensive network for for example, virtually all the product costs came from
accumulating and reporting costs at each of its more an allocation process.
than 5,000 cost centers. It summarized them by dif- Product cost estimates will not have the five- and
ferent classifications, geographical regions, and de- six-digit precision reported by a standard cost system.
No
grees of authority for all levels of management. By They will also be more subjective and less precise
comparing operating costs against budget and to the than the measurements in an operational control
same periods in the previous year, the system pro- system. Executives of multiproduct companies will
vided an excellent tool for cost control and produc- be fortunate if the first digit in their product cost
tivity improvement. estimates is valid, and they can make a reasonably
Then deregulation—and price competition—hit good guess at the second. But the estimates will
the company. Executives realized that none of the realistically approximate the long-run demands each
information in their elaborate reporting system could product makes on the organization’s resources.
help them to estimate product costs. Without knowl-
Do
edge of product costs, the new freedom to quote prices Cost Variability. A company should base most of
and to enter or leave markets could have been disas- its important product decisions on estimates of the
trous. Fortunately, the company developed com- long-run, variable costs of individual products.1
pletely new systems to estimate product costs and to Whether costs are fixed or variable, of course, depends
evaluate product and product-line profitability. The on the viewer’s time horizon. In the short run, virtu-
company is now prospering in its deregulated envi- ally all costs are fixed: materials have already been
ronment. acquired, utilities have been turned on, and the work-
Traditional standard cost systems in manufactur- ers have showed up for the day. Over a long period,
t
Product decisions have long-term consequences for production costs of today’s products.
os
the organization. Executives should therefore con-
sider virtually all costs to be variable when measuring Updates. A company does not need to perform the
product costs. That will require a new orientation for analysis and interviews for the product costing sys-
many managers. They must recognize that many tem more than once a year unless it makes major
costs traditionally thought of as fixed actually vary changes in its process technology, product mix, or
according to the diversity and complexity of prod- organizational structure. Decisions regarding prod-
uct introduction, abandonment, and pricing are stra-
rP
ucts. Much manufacturing overhead, for example,
comes from transactions associated with the start or tegic matters that should be based on the long-run
finish of production, such as placing and paying for marginal costs of each product.
orders, receiving and inspecting purchased materials, The annual product cost computation does not
setting up machines, moving inventory, and shipping have to be part of the main financial accounting
finished goods.2 system, nor does it require a lot of time and money
To reflect these costs, the system must include not to develop and implement. Several businesses have
only traditional volume-related measures for tracing developed prototype product cost systems on per-
yo
costs to products such as labor and machine hours or sonal computers. Of course, if a company has many
materials quantities, but also measures that count products going through complex production and dis-
setups, inspections, receipts, parts, vendors, and en- tribution processes, its product cost system will be
gineering change orders. The scheme must determine more expensive to build and operate.
how indirect production costs vary in the long run, Even with only annual updates, managers can use
both with regard to production volume and to the the system throughout the year to influence new
activities necessary to produce multiple items in the product design, introduction, and pricing decisions.
same facility. A good system yields unit costs for all key activities
op
(labor and machine hours, energy usage, materials,
System Scope. While the typical operational con- support), and it includes the unit costs of transactions
trol system segregates costs incurred at each respon- like setups, shipments, part and vendor quantities,
sibility center, a good product cost system should and inspections. A company can estimate a new
report expenses incurred across the organization’s product’s cost by specifying its demands on both
entire value chain. A product’s cost includes not only activities and transactions.
Including the costs of transactions like setups in
tC
All company resources support production and these features versus using simpler designs that ex-
sales. Even corporate expenses should be allocated to ploit existing parts and vendors.
product costs, especially if they vary across lines.
Legal expenses are a good example. They can vary by
risks of product liability and environmental damage, WHEN EASY OR DIFFICULT?
or by antitrust concerns across different categories of
products. One cannot generalize about the ease of designing
The product cost system can ignore only two adequate operational control and product cost sys-
classes of costs—expenses incurred that benefit fu- tems. Companies with only a single product can
Do
ture products, like basic research or development, estimate product costs with a trivial system. Accu-
and the expenses of idle or unused capacity. Existing mulate all the expenses during a period, subtract
financial accounting rules require that basic R&D be amounts relating to future products or excess capac-
expensed each period. But for managerial purposes, ity, and divide the remainder by the number of units
R&D should be considered investments in future produced. Similarly, companies with continuous-
products, not costs of present products. Unused ca- flow production processes that yield homogeneous
pacity is an expense for a particular period due to outputs can rely on measurement of product costs in
cyclical declines in sales, or an investment for future units, like cost per ton or cost per gallon. Product
t
machine is also rather simple. In contrast, it can be systems for operational control and product costing.
os
extremely difficult to estimate costs of items pro- Of course, an argument for expanding the number
duced by complex batch and assembly processes. of cost systems conflicts with a strongly ingrained
Operational control systems are simple to design financial culture to have only one measurement sys-
and install in highly repetitive production environ- tem for everyone. Eventually, designers may be smart
ments, especially those governed by well-understood enough to create such a system, but we don’t have
scientific relationships between inputs and outputs. one today. Any time accepted wisdom is overthrown,
rP
Operational control is also easier in functional or- the world suddenly looks far more complex. When
ganizations where each unit performs narrowly de- scientists declared a war on cancer more than a dec-
fined functions. Furthermore, an operational control ade ago, for example, they thought they would need
system can be installed inexpensively when produc- specialized cures for the hundreds of different forms
tion data are readily available. When product diver- of the disease. But over time and after extensive
sity is high, though, especially in production of experimentation, they have begun to develop unify-
unique items or with multiperiod production pro- ing theories that offer hope for more general treat-
cesses (as in construction, shipbuilding, and design ments and cures.
yo
and assembly of large, one-of-a-kind machines), op- In the same way, it is too early to discover the
erational control systems will be difficult to develop. general system that will meet all the organization’s
No single system can adequately answer the de- demands for cost information. Designers must first
mands made by the diverse functions of cost systems. attack the individual pieces, then with greater wis-
While companies can use one method to capture all dom and insight eventually discover a general cost
their detailed transactions data, the processing of this system that works for all managerial functions. Com-
information for diverse purposes and audiences de- panies that decide to wait for such a unifying dis-
mands separate, customized development. Compa- covery, though, will suffer in the interim the con-
op
nies that try to satisfy all the needs for cost informa- sequences of using inadequate information on
tion with a single system have discovered they can’t operating performance and product costs.
perform important managerial functions adequately.
Moreover, systems that work well for one company 1. See Robin Cooper and Robert S. Kaplan, “How Cost Account-
may fail in a different environment. Each company ing Systematically Distorts Product Costs,” in Accounting &
Management: Field Study Perspectives, ed. William J. Bruns, Jr.
has to design methods that make sense for its particu-
and Robert S. Kaplan (Cambridge: Harvard Business School Press,
tC
t
os
rP
HBR Subscriptions Harvard Business Review
U.S. and Canada Outside U.S. and Canada
Subscription Service Tower House
P.O. Box 52623 Sovereign Park
Boulder, CO 80322-2623 Lathkill Street
yo
Market Harborough
Leicestershire LE16 9EF
Telephone: (800) 274-3214 Telephone: 44-85-846-8888
Fax: (617) 496-8145 Fax: 44-85-843-4958
American Express, MasterCard, Visa accepted. Billing available.
HBS Management Productions Harvard Business School Management Productions videos are
Videos produced by award-winning documentary filmmakers. You’ll find
them lively, engaging, and informative.
HBR Custom Reprints Please inquire about HBR’s custom service for quantity discounts.
We will print your company’s logo on the cover of reprints,
collections, or books in black and white, two color, or four color. The
No
Permissions Editor
Harvard Business School Publishing
Box 230-5
60 Harvard Way
Boston, MA 02163
Telephone: (617) 495-6849
This document is authorized for educator review use only by Lenin Faican, Universidad Internacional SEK until Aug 2025. Copying or posting is an infringement of copyright.
[email protected] or 617.783.7860