Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Activity-Based Costing

at Hewlett-Packard's
Andover Surface
Mount Center
The major benefit from the ABC model was aset of
economic design heuristics for the R&D community

Karl Geffken

In the spring of 1987, Hewlett-Packard (HP) lations for both costing and performance mea-
hosted a seminar on a new cost accounting surement. John Browne, a manufacturing
technique called Activity-Based Costing (ABC). manager who was instrumental in our ABC
Several HP divisions had experience and model, remembers "We had a gulf between cost
agreed to share their knowledge with others in accounting information and actual incurred
the company. Interest in the topic was over- costs which would be useful for decision mak-
whelming. Acolleague of mine and I, aware of ing ... going to cost pools and drivers allowed
ABC before the seminar, were skeptical of this the physical process to be modeled closer to
latest fad. By the time we got back to our reality."
home division, we envisioned its usefulness as Surface mount manufacturing is a capi-
a measurement tool. tal intensive environment where direct labor
The timing was perfect; we were in the consists of no more than four percent of the
process of developing a new surface mount total product cost. Continued use of the tradi-
manufacturing line. We sold key managers on tional method would add no value to the organ-
ABC, and were off developing it for the new ization and would probably provide misleading
organization. We introduced our ABC model at standard costs and decision-making informa-
the beginning of FY (fiscal year) 1988; four tion.
years later, it's become an integral decision Further, the only decision support tool
making tool for the now well-established from the cost accounting system was a labor
Andover Surface Mount Center (ASMC) in variance report which told managers that if
Andover, MA. We've had ups and downs, twists they overspent direct labor by $5000, then
and turns, but we've never for a moment overhead would be overabsorbed by $20,000.
regretted dumping our traditional labor based This concept of absorption was not useful to
system in favor of ABC. manufacturing managers, let alone the cost
IJIssatIsfat:tJlHI with Labor-based System accounting folks. It reinforced the image of
The real push toward ABC came from a cost accountants as bean counters wearing
general dissatisfaction with our traditional green eye shades, hunched over ledgers in the
labor-based SYStem: it was not meeting expec- backroom.
15
March / April 1992
ASMC Average Board Cost by Category

SMT Auto-load 5.0%

Post-wave Hand 4.0%

Panel 3.0%

Tes! I Repair 3.0%

Procurement OH 3.0%
EVlliutilln lit the Mlldel
Pre-wave Hand 2.0%
The ABC model of four years ago bears little resem-
blance to what we use today. ABC models evolve as one
learns more about their manufacturing process and
gains confidence using the technique.
In 1988, the first full year of production, our ABC
model consisted of two drivers and cost pools: auto-load
SMT parts and hand-load parts. This directly captured a
Figure 1.
majority of the activity while highlighting the mix of part
types on each board.
For the first time, we had solid fact-based informa-
Evolution of ABC Model at ASMC - Production Cost Drivers
tion revealing major cost differentials: insertion of hand-
T851 + repair lime
Panels
load parts cost ten times more than auto-load parts.
Panels ------·1 Auto-loaos
"This blew quite a few people away," commented a team
Hirld-load placements Pre-wave hand-IDIIlls
Investigalion Auto-load placemenls Post-wave lalli-loads member. From that time on, the model was accepted and
respected throughout R&D, manufacturing, and finance
"87 1988 1989 1990 1~1 1~2
as a major decision support tool. (See Figure 2.)
By 1990, as ASMC's business expanded, complexity
expanded as well. We knew a large portion of costs was
Figure 2. not associated with the two existing driver pools but
Keep It Simple
instead was focused around panels of boards flOWing
through the process. Panel-driven activities included sol-
The Andover Surface Mount Center is a what HP
der paste deposition, paste reflow, wave solder, defluxing
calls a "high mix, medium volume" board assembly
and de-panel. We wanted to communicate a flat rate per
operation. We build over 100 different assemblies for sev-
panel and then let the customer decide how to best use
eral internal customers. The boards are then shipped to
the panel space. Apanel cost driver was then introduced.
customers and assembled into electronic medical equip-
Heading into 1992, ASMC production managers
ment such as ultrasonic imaging products and operating
became acutely aware of the process cycle times and
room monitors.
wanted further refinement of the model to accurately
The first step in developing the model was to under-
reflect the cost of boards that Zipped through the line ver-
stand, from manufacturing's perspective, their vision for sus ones that were consistently slower. We noticed that the
the new surface mount center. Out of the discussions we slower boards got hung up in two tail-end processes:
developed a good sense of the key processes, tools needed hand-loading of parts after the wave solder and
to better evaluate performance, and overall operating lesVrepair. Additionally, we now recognized that these two
philosophy. processes made up over 35 percent of the overhead spend-
The point that influenced us most was manufactur- ing!
ing's insistence to redesign/re-engineer work to minimize We modified the model once again to highlight
process costs. Their philosophical slogan was to "Keep It these high cost areas. By dividing hand-load components
Simple." The new surface mount center would be a lean into two groups, parts wave-soldered versus parts hand-
and mean operation where complex work-flows and lay- soldered, an astonishing 20 times difference in cost
ers of bureaucracy would not be tolerated. "Keep It Sim- appeared. The production manager John Cadigan noted,
ple" became the backbone of our model as we focused "You could see where lots of people were doing hand-sol-
only on major activities which drove the spending in dering. It had to cost more."
ASMC. "Keep It Simple" made absolute sense to us, espe- ASMC's customer base had mixed reactions to the
cially since overhead was only 20 percent of the total latest change. Some were pleased to no longer subsidize
board cost. (See Figure 1.) slower boards: others complained about paying signifi-
16
Target
Andover Surlace Mounl Center Production Process Flow, Cost Drivers, and Activity Pools

Pasle
deposit f---- SM!
loading f---- Poste
rellow - Hand
IMding - Wd~e·
solder f---- Dellux
"d
clean
- KaM load
and solder f---- Depanel
- Te51
repair
r----o Silip

Pre-wave
r.ut Pool Panel Process SMT Loading Panel Process Hand-Loadng Panel Process Panel Process Post-wave 'allel Process TestJRepair
Hard IC-Jbing

, Pre-wave I Post-wave
Driver HLParts I PallfJls I Panels HL Parts f ""Is TJRtlnil

Figure 3.

cantly more for boards that haven't changed in several I. The insertion cost of through-hole parts which are
years. Customers voiced a common fear that the model hand-soldered is 60 times the cost of SMT parts.
was growing too complex and changing too often. 2. The insertion cost of through-hole parts which are
We explained to our customers that our philosophy wave-soldered is 5times the cost of SMT parts.
of "Keep It Simple" was alive and well. Afive-cost driver
3. Test and repair time is $X per minute.
model was still relatively simple, easy to maintain and
communicate. The benefits of a more detailed model 4. Each panel cost $X to process reganlless of the number
were worth the effort since it exposed some critical cost of boards on it.
differences in the manufacturing process. (See Figure 3.) 5. Material, representing 80 percent of product cost, is the
biggest opportunity area.
Benefits
Since 1988, the model has successfully impacted These economic heuristics have caused new designs
many areas of our organization. Specifically, we've seen; to be lower cost. Customers are also revising or redesign-
ing boards to reduce through-hole content and to better
• Benefits to R&D by providing economic design heuris-
utilize panel space.
tics
The expansion of our model to five drivers has
• Improved visibility of process costs enhanced its usefulness as a credible DFCA (designing for
• Improved performance metrics competitive advantage) tool. Customers are keenly aware
• Improved communication of the financial consequences associated with boards that
are difficult to manufacture. (See Figure 4.)
• Improved product costing.
Improved Visibility of Process Costs
R&D Impact
By separating process areas such as hand-loading
The major benefit from the ABC model was a set of
after wave solder, ASMC management has improved visi-
economic design heuristics for the R&D community.
bility of spending patterns by distinct activities. High unit
Prior to the model, the only way to assess overhead on a
cost activities stick out like red beacons and become
new product was to estimate the labor content and multi-
prime targets for TQM, productivity improvement, or cost
ply it by the average overhead rate. Everyone was street-
reduction programs.
wise enough to know that reductions in direct labor
didn't automatically translate to decreases in other plant Improved Performance Metrics
operating costs. Most engineers were happy to finally More importantly, the ABC model has proVided key
have a tool which gave them direct cause-and-effect rela- financial performance metrics for ASMC. People can
tionships to the black hole of overhead. Now engineers wrap their arms around a unit cost easily, using it to
ask questions about driver rates; some want small ver- assess overall performance. Unit costs derived through
sions of the model so they can construct ''what if' scenar- ABC do have a tendency to simplify issues, but overall it's
ios on their own. a metric that everyone from the shop worker to the high-
Our goal was to create a short list of key driver rates est manager can identify with. Many top-level managers
that the R&D community would internalize and use are keenly aware of these driver costs and, as a result,
automatically when making design decisions. Although many strategic and tactical decisions are made based on
the list of driver rates has grown over the past several them.
years, it is still a relatively small set. Improved Communication
ASMC's design rules focus on five main items. The model has helped the finance organization
17
MarchiAprl/1992
Fictitious Example of Board Revision and Cost Impact
Board characteristics Original design Revised design

Number of SMT loaded parts 300 305


Number of pre-wave hand loads 10 15
Number of post-wave hand loads 15 5 1. 7echnology limits. In my example, hand-loads have
Number of boards per panel 6 6
Average test/repair minutes 3 2
not disappeared because our products require many
Material cost $500.00 $500.00 analog parts that are not available in SMT packages.
-_.--
2. Priorities. Design engineers must balance require-
Total board cost $636.50 $589.40
ments other than ODSt improvement. Time-to-market,
Driver rates' functionality, and high quality all bid for an engi-
SMT parts $0.08 neer's resources. Financial considerations may get
Pre-wave parts $0.30 subordinated when overhead is only 20 percent of the
Post-wave parts $4.50 board cost.
Per panel" $30.00 3. Redesigning is expensive and hard to recoup on
T/R Cost (per minute) $4.00
Material overhead rate 5.00% low-volume boards. ASMC builds many low-volume
boards due to the nature of the finished product, med-
'Rales have been modified for the example. ical equipment. If you build high-volume consumer
" Pasle deposition. paste reflow. wave solder, deflux. depanei. products, redesign may be worthwhile since your pay-
back time is shorter on your investment.
Figure4.
An ABC model by itself will not initiate major
gain credibility and become a more active business part- changes in product design or processes. There are many
ner. 1\vo-way oommunication between finance and man- other factors at play. Your expectation level of what to
ufacturingIR&D has improved, due in part to finance achieve from ABC must be realistic.
fotks spending a great deal of time on the line under- Simple ABC Models Hare Limitations
standing the processes. As part of the model's develop- Our model has been kept relatively simple in order
ment, they were located with a desk and a PC right in the to focus in on key activities only and to minimize the
middle of the production area. "Under these conditions model's upkeep on an ongoing basis. The result is a tool
you couldn't help but feel like you're helping to put the which gives us great decision information for most but
boards together," commented a team member from not all, production situations. For example, we haven't
finance. separated out the cost of bending or clipping legs on
Accurate Product Costing components. Although it would be nice to know the cost
To a lesser extent, but still an important benefit, of those activities, they are minor processes and not worth
our ABC model has helped distribute overhead costs more the effort to track.
fairly to the various system product lines. To those work- Arule of thumb ooined by Tom Elliot, an HP oorpo-
ing on low-volume products our input is critical to their rate financial project manager, is that "the value of the
project's success. Using our driver rates, dozens of boards cost driver information must be greater than the ODSt of
have been tuned to get a new product's manufacturing providing it,"
cost down to a targeted level. ABC Is Not a PanactNI
What Has Not Come Out of the Model ABC is not a panacea or Holy Grail which will auto-
Reality fell short of personal expectations. It was matically make organizations that use it more competi-
my personal expectation several years ago that continued tive. Attention to cycle time, flexible manufactUring, sim-
use of ABC would drive out most high unit ODSt elements plified work-flows, and common sense contribute more to
of boards. For example, with hand-loads costing many success than ABC. The truth is that ABC only highlights
times more than auto-loads, it would seem to be a matter opportunities. Using ABC is easy; the hard part is making
of time before boards would contain very few, if any, the right management decisions to change the process.
manual parts. That has not happened. Questionabl. use of ABC in other areas
Overall, improvements have been made to board Suggestions that ABC is applicable for processes in
designs but ABC has not radically changed the engineer's marketing, sales, R&D and administration are common.
design process. Here are some possible reasons: Theoretically, it makes sense because it's such a benefi-

18
Target
cial tool. In reality, it hasn't taken off with gusto at our About Helllrletl:·Packalrd
site in areas other than manufacturing. I'm theorizing Hewlett·Packard is an intematiot:!ar!llllttdiilctGhji
why it's the case, but I think that people still see it funda- of measurement and computational products and
mentally as another accounting control mechanism. The systems recognized for excellence and quality in
very name "Activity-Based Costing" turns many people support. The company's products and services
off. are used in industry, businesses, engineering,
Everything is marketing. Perhaps acceptance of science, medicine, and education in
ABC would be more widespread if the name of it were approximately 100 countries. HP has 89,700
employees and had $14.5 billion in revenue for
changed to "Activity-Based Decision Making!"
1991.
Concluding 1IBmattrs: LolJ1clng Sat:1r at Four Years
• Before any organization embarks on ABC, there must • Walk before you run. When developiog your ABC
be absolute commitment from the manufacturing model, start out with a simple set of drivers. Expand
and R&D functions that the model will be used, sup- as you more fully understand your customers' needs,
ported, and owned by them. Finance should only be a the manufacturing process, and the administration of
f~ilitator. With cross-functional inputs, chances are the model.
the model will be richer in content and not viewed as • Keep in mind that ABC is a decision support tool first
just another accounting tool. and an accounting tool second.
• The level of complexity in the model must be clearly
Karl Gef/ken is the finance manager for HP's Medical
understood before it is implemented. The model ManufactUring operations in Andover, M4. He holds an M.B.A.
should be well suited for the job. Let the needs and from /J!bip,b, ({lid (/ B.A, in economksfrom Muhlenberg College.
desires of the customers who use the information
drive the model's outcome. Also, the model's complex-
ity may be directly correlated to the degree of over-
head in the total product cost. Products with low over-
©1992AME
head don't require a very detailed ABC model. For infonnation on reprints, contact:
• Market the ABC concept to everyone that the model Association for Manufacturing Excellence
380 West Palatine Road
impacts. People who are in the mainstream will have Wheelin~ IL 60090
little trouble bUying into it, but there will be skeptics 70&'520-3282
who view this as more cost accounting hocus pocus.
~
Education and communication are key to its success.
Even when revisions are made to the model, get your
customers' buy-in before implementation. They will
appreciate the opportunity to provide feedback.
• Developing and perfecting an ABC model is an evolu-
tionary process. Changes to the drivers and activity
pools may happen as one gains experience. The
important point is to not make too many modifica-
tions over a short period of time. Frequent changes
will confuse the intended audience, make period to
period comparisons difficult, and the model's facilita-
tors may lose Credibility. Stability is desirable because
the model will reinforce decisions that will have a
long-term impact.

19
March / April 1992

You might also like