Operations Management Assignment

ECONOMY

by Gianpaolo Callioni, Xavier de Montgros, Regine Slagmulder, Luk N. Van Wassenhove, and Linda

Wright

FROM THE MARCH 2005 ISSUE

he 1990s were hard for the PC business. Although demand grew �vefold between 1990 and

1997, the products had become household staples, and it was di cult for companies to

di!erentiate their o!erings.

Hewlett-Packard came out better than many. The computer hardware giant slashed prices on all of

its PCs: 10% in 1991, another 26% in 1992, and yet another 22% in 1993. At the same time, it

revamped its design, planning, and production processes to shorten cycle times, respond quickly to

changes in demand, and move inventory to the right location as it was needed. By late 1999, HP had

displaced IBM as the world’s third-largest PC manufacturer in terms of revenue, behind Dell and

Compaq.

But for all its success in maintaining market share, HP was struggling to turn a dollar. By 1997,

margins on its PCs were as thin as a silicon wafer, and some product lines had not turned a pro�t

since 1993. Price cuts made formerly insigni�cant costs critical—computer manufacturers simply

could not stock up on components or any other inventory. Any excess at the end of a product’s short

life had to be written o!, further eroding margins. Adding to the pressure, constant technology

advances made new products obsolete in as few as six months. A common rule of thumb was that

the value of a fully assembled PC decreased at the rate of 1% a week. Although HP’s supply chains

were 4exible and responsive enough to deliver the PCs when and where customers wanted them,

they were not economically sustainable.

Compounding the problem, company executives realized, was the fact that HP’s management-

Inventory-Driven Costs https://hbr.org/2005/03/inventory-driven-costs#

1 of 13 10/7/2015 5:29 PM

accounting metrics had failed to keep pace with the evolution of its supply chains. HP now oversaw

a complex, multitiered manufacturing network made up of many disparate entities. But the

company’s current approach to cost measurement allowed individual players in the chain to see

only their piece of the puzzle, making it impossible for them to assess the overall dollar impact of

their local decisions. They could not see the e!ects of operational decisions like setting the size of

bu!er inventory between two sections of the production process. Nor could they assess the impact

of such strategic decisions as choosing to locate a �nal assembly plant close to a particular supplier.

If the company wanted to design sustainable supply chains, it needed to give all its managers a

direct line of sight to the overall bottom line.

The Hidden Cost of Inventory

To make the PC business more cost competitive, HP’s Strategic Planning and Modeling (SPaM)

group, led by Corey Billington, undertook an exhaustive review of the PC business’s overall cost

structure in 1997. It soon became clear that mismatches between demand and supply leading to

excess inventory were themain drivers of PC costs; in 1995, for example, costs related to inventory

had equaled the PC business’s total operating margin. It was just as clear that the division’s existing

cost metrics did not track all of HP’s inventory-driven costs (IDC), pieces of which were often mixed

in with other cost items, scattered over di!erent functions and geographic locations, and recorded

at di!erent times using di!erent accounting conventions.

The most readily identi�able component of inventory-driven costs is the traditional inventory cost

item, usually de�ned as the “holding cost of inventory,” which covers both the capital cost of money

tied up in inventory and the physical costs of having inventory (warehouse space costs, storage

taxes, insurance, rework, breakage, spoilage). At HP, however, the holding cost accounted for less

than 10% of total inventory-driven costs. SPaM’s investigation revealed four other inventory-driven

cost items at HP’s PC business. And each of them needed to be managed in a distinct way.

Excess inventory was the main driver of Hewlett-Packard’s PC costs; one year, in fact, inventory-driven costs equaled the PC business’s total operating margin.

Inventory-Driven Costs https://hbr.org/2005/03/inventory-driven-costs#

2 of 13 10/7/2015 5:29 PM

Component Devaluation Costs.

According to SPaM’s calculations, these accounted for the lion’s share of HP’s inventory costs. Key

components such as microprocessor chips and memory typically drop in price quickly and steeply.

The price of a CPU, for instance, might fall as much as 40% during its nine-month life cycle, and the

penalties for holding excess parts when a price drop occurred could be enormous. Back in 1997,

however, few electronic hardware makers had realized just how perishable their goods had become,

and HP, in common with many others, maintained inventory in several places—in factories and

distribution centers, in merge centers, in transit. Every time the component prices fell, HP was hit

with another devaluation cost at each of these points in the value chain.

HP had no control over component prices, but it could control how much inventory it was holding.

That meant reducing the number of nodes in the supply chain, consolidating manufacturing

facilities, taking possession of components on a just-in-time basis, paying the going price at that

time, and working with suppliers to minimize inventory when a price drop was anticipated.

Price Protection Costs.

If HP dropped the market price of a product after units had already been shipped to a sales channel,

it had to reimburse its channel partners for the di!erence for any units that had not yet sold, so the

channel partner didn’t have to sell at a loss. Given how quickly value decayed, this mismatched

inventory exposed HP to big price protection risks. A channel partner might buy a product from HP

when the prevailing market price was $1,000. But if the item sold �ve weeks later at a new price of

$950, HP had to reimburse the $50 di!erence. To limit this cost, HP had to be certain that channel

partners’ inventory never exceeded the minimum number of days required to ensure the desired

availability, so that no excess inventory had to be protected.This meant that Hewlett-Packard had to

keep its manufacturing turnaround times short and replenishment cycles frequent. HP also o!ered

its channel partners incentives to carry the lower levels of inventory. Analysis showed that the cost

of these incentives was almost always lower than the cost of reimbursing channel partners after

price breaks.

Product Return Costs.

In a sense, product return costs are simply 100% price protection costs; distributors can simply

return unsold goods to the manufacturer for a full refund. In some cases, product returns

constituted more than 10% of the product’s revenue, not because of product failures but because

Inventory-Driven Costs https://hbr.org/2005/03/inventory-driven-costs#

3 of 13 10/7/2015 5:29 PM

resellers were returning excess inventory. Apart from incurring the operational costs (shipping,

handling, product retesting, and the like), returns lengthened the time a product spent in the supply

chain before reaching an end user, increasing HP’s exposure to additional devaluation risks and

inventory �nance costs. To manage this type of inventory-driven cost, HP had to work closely with

channel partners to optimize the whole supply chain. By agreeing with its channel partners on

speci�c inventory levels and delivery expectations, HP reduced ine cient inventory in the channel

and increased its overall quality of service both to partners and end customers.

Obsolescence Costs.

End-of-life write-o!s were initially the most obvious portion of this cost. PC life cycles being so

short, even a small miscalculation in anticipated demand could leave the company holding stacks of

worthless goods that had to be written o!. The other related but sometimes less obvious

components of obsolescence costs were discounts on about-to-be-discontinued products and the

associated marketing e!ort required to accelerate their sale. These costs are typically not included

in a company’s cost of goods sold and, although HP executives took these discounts and marketing

costs into account when making decisions about discontinuing product lines, they seldom

considered them when determining the real cost of inventory. To avoid obsolescence costs, HP had

to be very e cient in managing product introductions, so that new models were launched just as

the last remaining units of the old models sold out.

On the whole, calculating these components of inventory costs is fairly straightforward. The

simplest to determine is devaluation, which can be �gured by multiplying the inventory level of the

product or component in question by the proper devaluation rate. Suppose that a hypothetical

company sells a consumer electronic item that devalues at a yearly rate of 60%. Average annual

inventory related to the product is worth $200 million, and annual revenues from selling the

product are $1 billion. Then, the yearly inventory-related cost due to product devaluation is the

average annual inventory multiplied by the devaluation rate, or $120 million. Dividing that �gure by

revenues gives us the cost as a margin, in this case 12%. Clearly, at any given devaluation rate,

inventory costs will increase directly as inventory increases. The faster prices fall, the more the

inventory-driven costs rise as inventory increases. Since in most cases the devaluation rate is

outside managerial control, the only way to reduce the impact of devaluation on pro�ts is to do a

better job of matching demand and supply, thereby shrinking inventories.

Inventory-Driven Costs https://hbr.org/2005/03/inventory-driven-costs#

4 of 13 10/7/2015 5:29 PM

Tracing Costs to the Source

Price protection and return costs are calculated in a similar way, but the actual sums are somewhat

more complicated because they depend on the contractual agreements set up between

manufacturers and distributors. In its simplest case, if a manufacturer has agreed to reimburse its

distributors 100% whenever it lowers its list price, the formula for determining price protection

costs is the price drop times the number of units of the product in distributors’ inventory. Similarly,

return costs would be the number of items returned of a particular product times the wholesale

price paid by the retailer in the �rst place. But those sums need to be adjusted by the contract terms,

which might, for example, not allow all inventory to be returned if the retailer deliberately

overstocks. (The more generous the price-protection and return terms, the less the distributor will

lose by doing so.) What’s more, contract enforcement may depend on speci�c circumstances. For

example, even if the contract constrains the return options of the channel partner, HP may still

agree to take products back to maintain good relations and secure future sales.

Devaluation, price protection, and return costs are essentially continuous costs; they occur all the

time and can be calculated at any point. Obsolescence costs, however, are discrete, arising only

when a company decides to retire a particular product and therefore cannot be estimated until that

moment. The amount of obsolescence costs is determined by several factors. First, the company

needs to write o! 100% of the value of �nished goods in its inventories (less any recycling or scrap

bene�ts). Then it must write down the value of any components in the pipeline. If components are

product speci�c, they will have to be written o! 100%; those that can be used elsewhere will be

subject to devaluation costs depending on how quickly they can be transferred to other products.

Finally, the company has to add in the related marketing and discounting costs of selling o!

about-to-be discontinued products in �re sales.

As important as identifying the various hidden components of inventory costs was to HP, even more

powerful was understanding how the impact of each IDC component di!ered for di!erent products.

That had profound implications for the way HP managed its product portfolio. This insight is

illustrated in the table “Tracing Costs to the Source,” which compares three unnamed HP products.

Total IDC is relatively high for product A, at 14.25% of product revenues, and fully half of this total

comes from price protection costs. Total IDC for product C, by contrast, is only 4.80% of revenues,

and the largest share of that is component devaluation.

These �gures suggest how the supply chains of

Inventory-Driven Costs https://hbr.org/2005/03/inventory-driven-costs#

5 of 13 10/7/2015 5:29 PM

When the components of inventory costs are

broken out, it becomes easy to see how the supply

chains of different products need to be adjusted

in different ways to lower costs at their source. In

this example of three of Hewlett-Packard’s

products, the highest costs for product A are

coming from goods whose prices have dropped

after they’ve been shipped to retailers. But the

greatest problem for products B and C are drops

in component prices before the products ever get

out of the factory.

the three products need to be managed. Since the

inventories of channel partners represent the

largest component of inventory costs for product

A, what managers need to do is improve supply

chain management downstream: They must do a

better job of forecasting demand. They need to

encourage vendor-managed inventory (VMI) and

collaborative planning, forecasting, and

replenishment (CPFR) initiatives. By contrast,

products B and C probably need better upstream

management with suppliers or with product

designers to reduce component devaluation risks.

The four cost items that SPaM identi�ed are not

the only kinds of potential inventory-driven costs.

Companies that maintain high stocks of raw

materials, for instance, may well �nd that the write-down in the value of their inventory, stemming

from reductions in raw material prices, can in any one year outweigh the bene�ts of the lower input

prices. Another common source of IDC are price discounts, which typically arise when errors in

forecasting demand lead to excess inventory that the company is forced to sell at below-market

prices, resulting in lower margins on the products in question.

The Turnaround

The Mobile Computing Division (MCD) was the �rst of Hewlett-Packard’s PC units to take inventory-

driven costs into account when formulating strategy. In 1998, before redesigning its supply chain,

MCD was taping dollars to every machine it shipped. Year after year, its managers tried di!erent

initiatives to improve pro�ts, including reducing the cost of materials, controlling operating

expenses, and generating revenue growth through new product development. Nothing worked until

the notebook division began to look at the impact of its supply chain decisions on IDC.

Intuitively, MCD’s managers believed that the unit could not become pro�table until it consolidated

all worldwide manufacturing at a single location and shipped the �nished products directly to

customers. This would, however, represent a major strategic change, and the division’s executives

Inventory-Driven Costs https://hbr.org/2005/03/inventory-driven-costs#

6 of 13 10/7/2015 5:29 PM

Finding the Lowest-Cost Option

Breaking out inventory-driven costs gave HP’s

Mobile Computing Division a far more

comprehensive understanding of the various

supply chain options it was considering. Scenario

1 represents the existing cost structure of the

unit’s business in 1998, in which products were

manufactured centrally and configured to

conform to local needs in regional markets.

Scenario 5 is the centralized one-step

manufacturing model that HP eventually

adopted. Its merits do not become apparent until

inventory-driven costs are taken into account.

worried that centralized manufacturing would a!ect the unit’s ability to give customers the service

they wanted, resulting in the loss of both market share and revenue. To build a compelling case for

making the move, they needed to quantify the risk—and the opportunity—involved in making such a

radical change.

Collaborating with SPaM, the MCD team considered many di!erent supply chain scenarios. The

exhibit “Finding the Lowest-Cost Option” compares �ve of these. For reasons of con�dentiality, the

actual numbers have been disguised, but the exhibit truthfully reveals the relative cost di!erences.

Scenario 1 represents the original con�guration, a two-step supply chain with a central

manufacturing facility and a certain amount of local product con�guration carried out at regional

facilities. It turns out that about 40% of the true total supply chain cost in this scenario is linked to

inventory. The company had never taken inventory costs such as devaluation and obsolescence fully

into account before, so the exercise revealed just how expensive its past supply chain decisions had

actually been.

Scenarios 2 and 3 essentially retain the basic

two-step supply chain con�guration of scenario 1

but assume that the company will limit inventory

costs by putting more-e cient manufacturing

processes in place and by passing on to supply

chain partners as much responsibility as possible

for inventories. As the exhibit shows, these

nonstrategic changes would su ce to reduce total

costs (including IDC) by as much as 24%. In

scenarios 4 and 5, the unit considered radically

restructuring its supply chain, and either

approach would involve major strategic changes

to implement. Both are essentially one-step

supply chains, but there the similarities end.

Scenario 4 assumes that HP would have several

regional factories doing both manufacturing and

localized con�gurations. In scenario 5, however,

everything would be done in a single central

Inventory-Driven Costs https://hbr.org/2005/03/inventory-driven-costs#

7 of 13 10/7/2015 5:29 PM

factory and airfreighted directly to customers

worldwide. What this means is that scenario 4 has

higher manufacturing costs than scenario 5 but lower freight costs.

How should the Mobile Computing Division choose between the scenarios? In terms of

manufacturing costs only, scenario 5, which MCD called International Direct Ship, is the best option,

suggesting that the company should consider making the radical change to a one-step supply chain.

Factoring in freight, the one-step model remains the theoretically optimal supply chain, although

scenario 4, with regional factories, now appears to be the best option. But the case for switching to a

one-step model is considerably undermined because scenario 3, the best-case two-step option, is

now only one percentage point more expensive than scenario 4 and actually cheaper than scenario

5. The one percentage point di!erence is not enough to make up for the organizational and �nancial

changes that switching to a one-step supply chain would involve, despite the tight margins on the

PC business.

But what a di!erent picture emerges when IDCs such as inventory devaluation and obsolescence

costs are fully quanti�ed and included in the analysis. Then, the case for switching to a one-step

supply chain looks much stronger. Even if the company were to exploit all the possible opportunities

for reducing IDC in the current supply chain structure, as scenario 3 rather unrealistically assumes,

both one-step scenarios come out about four percentage points cheaper, which is enough to justify

management’s case for switching to a one-step supply chain. A full assessment of IDC, therefore, not

only revealed the limitations of MCD’s current supply chain but also provided strong evidence that

the only way the division could turn its business around was by refashioning its entire supply chain

strategy.

IDC considerations also drove the choice between the unit’s two very di!erent one-step supply

chain alternatives. As the exhibit shows, the total cost di!erence between them, when IDCs were

taken into account, was minimal (0.2 percentage points). But managers felt that scenario 4 o!ered

many opportunities for losing control of inventory-driven costs, and it was not clear that MCD would

always be able to contain those risks. Managing a multitude of regional factories with separate pools

of materials and products is harder than managing just the one. More factories would mean greater

chances that capacity would fall idle; several plants would certainly require higher levels of capital

investment than a single one. And when MCD introduced new products, several factories would

Inventory-Driven Costs https://hbr.org/2005/03/inventory-driven-costs#

8 of 13 10/7/2015 5:29 PM

have to ramp up their production processes, not just one. So, although scenarios 4 and 5 seemed to

be very similar in terms of base-case total costs, the Mobile Computing Division eventually plumped

for scenario 5.

MCD started reaping the bene�ts almost at once. Inventory-driven costs dropped from 18.7% of

total revenue in 1997 to 12.2% in 1998. In 1999, the division saw an even more dramatic

improvement, as IDC dropped to a mere 3.8% of revenue. These reductions translated directly into

savings for the notebook division’s bottom line, which in 1998 broke even and in 1999 turned

pro�table. The IDC metrics enabled the Mobile Computing Division to clearly identify exactly where

changes to its supply chain were needed and to justify changes that, according to more traditional

thinking, would not have made sense.

Other PC units, notably the commercial desktop business, started to follow MCD’s approach to

costing in their supply chain decisions, with similarly impressive results. Inspired by these

successes, HP decided to o cially implement the new metrics in all of its PC operations. In the �rst

stage of the rollout, the focus was on tracking IDC across all divisions. In some cases, the

information had to be assembled manually; in others, it was possible to automate the data

gathering. In the next stage, goals for each IDC line item were set for each region and product line,

based on projects that were known to be in the works and estimates of future market conditions. A

consistent presentation of the data was almost as important as the actual data. HP’s regional units,

product divisions, and �nance groups worked closely together to develop a standard template for

supply chain metrics, making sure that the data could be collected in the requested format for each

line item in a timely manner. So everyone uses the same line items on their spreadsheets, each with

the same de�nition, the same source of information, and the same method of calculation. As a

result, all users can look at a line item and know exactly where the data came from and what they

really mean.

The Payoff

HP can now manage the pro�tability of its value chain in a much more sophisticated way. Gone are

the days of across-the-board measures like “Everyone must cut inventories by 20% by the end of

the year,” which would usually result in a sometimes counterproductive 4urry of cookie-cutter lean

production and just-in-time initiatives. Now each product group is free to choose the supply chain

con�guration that best suits its needs as long as it meets the global IDC target. Product group

Inventory-Driven Costs https://hbr.org/2005/03/inventory-driven-costs#

9 of 13 10/7/2015 5:29 PM

managers may well have known before, on an intuitive level, what they needed to do, but the IDC

metrics have made it easier for them to convince senior managers that their particular situations

require particular solutions.

Incorporating the IDC metrics into decision making has also saved managers from moves that make

perfect sense for their own unit but add to overall costs. Previously, for example, a manager might

have decided against shipping goods by air because the extra transportation costs would have

exceeded the identi�able cost of �nancing and warehousing local inventory. But that decision would

have imposed costs elsewhere in the supply chain, which might well have exceeded the extra

transportation costs. Without measuring total IDC, there was no way to know that, and even if the

manager made the right decision, he would probably have been penalized for it. Now, however, he

would be rewarded for incurring extra local cost in the interests of reducing total costs.

The IDC metrics are valuable in a whole range of R&D and marketing decisions. Many downstream

supply chain costs arise because of choices managers make upstream in the product design phase.

The IDC discipline has made HP’s product designers much more aware of the consequences of their

decisions, which makes them more responsible and accountable. Before, someone who had

speci�ed a hard drive that took three months to obtain would probably not have realized that during

those months HP was liable for excess inventory, devaluation, and obsolescence costs. Now, such a

designer no longer has that excuse. The IDC metrics also help managers decide how much 4exibility

to build into new products. In the past, HP often underestimated the related supply chain costs of

o!ering lots of product features. Being able to quantify the real inventory-driven cost of adding, say,

Lithuanian language customization to a product helps in determining whether or not to o!er

customers that option.

But perhaps the greatest bene�t of the IDC metrics is that they link operational decisions to

corporate goals for creating shareholder value. In the new pro�t-focused climate, HP has been

abandoning its traditional �nancial-performance metric of return on sales in favor of return on net

assets (RONA). This re4ects the competitive reality that, for companies like HP, advantage derives

less from market share than from how e ciently the �rm manages its assets—in other words, its

supply chain. As the exhibit “Linking Inventory Costs to Financial Performance” shows, the

relationship between inventory-driven costs and return on net assets is direct, simple, and powerful,

which makes it far easier to align the interests and decisions of managers up and down the

Inventory-Driven Costs https://hbr.org/2005/03/inventory-driven-costs#

10 of 13 10/7/2015 5:29 PM

Linking Inventory Costs to Financial Performance

Hewlett-Packard is finding that return on net

assets (RONA) is a more accurate measure of

shareholder value than market share, because in

such price-sensitive industries, the key to

financial health is not revenue growth but sound

asset management. To measure RONA accurately,

companies need to track more than just

traditional inventory costs. Those costs affect

expenses only, but the price protection and

product return costs that inventory-driven cost

metrics track can also erode revenues. What’s

more, lowering inventory-driven costs can not

only decrease total costs and raise revenues but

also lower working capital requirements by

reducing the number of days of inventory

outstanding.

hierarchy.

The �nancial bene�ts have come quickly. HP’s

Personal Systems Group, for example, saw

worldwide inventory decline by 50% between

2000 and 2002, and it has maintained that level

ever since. Costs associated with inventory have

dropped even further, by some 70%. Since HP’s

merger with Compaq in May 2002, the push to

adopt IDC companywide has moved forward. At

this point, all of HP has adopted a standard set of

inventory-driven cost metrics.• • •

Hewlett-Packard isn’t the only company, of

course, that operates in a dynamic, highly price-

competitive industry. Consumer electronics,

fashion producers, and fresh-goods retailers all

face similar challenges. Any company with low

margins, short life cycles, highly perishable or

seasonal products, and unpredictable demand

needs to track the various components of its

inventory-driven costs. Without appropriate

performance metrics to help visualize the

magnitude of their supply chain problems and to

prompt people to take action, these �rms will simply not know if they are leaving hefty piles of

money on the table.

A version of this article appeared in the March 2005 issue of Harvard Business Review.

Gianpaolo Callioni (gianpaolo_callioni@hp.com) is the director of supply chain strategy and planning at Hewlett-Packard in

Palo Alto, California.

Inventory-Driven Costs https://hbr.org/2005/03/inventory-driven-costs#

11 of 13 10/7/2015 5:29 PM

Xavier de Montgros (xavier_de-montgros@hp.com) is the supply chain development director of HP’s Personal Systems

Group.

Regine Slagmulder (regine.slagmulder@insead.edu) is an associate professor at Insead in Fontainebleau, France.

Luk N. Van Wassenhove is the Henry Ford Chair of Manufacturing at INSEAD and director of the Humanitarian

Research Group. He is also an Academic Fellow at the European Center for Executive Development (CEDEP).

Linda Wright (linda_wright@hp.com) is a finance manager in HP’s Personal Systems Group.

Related Topics: OPERATIONS MANAGEMENT | SUPPLY CHAIN

This article is about ECONOMY

 FOLLOW THIS TOPIC

Comments

POS T

0 COMMENTS

POSTING GUIDELINES

We hope the conversations that take place on HBR.org will be energetic, constructive, and thought-provoking. To comment, readers must sign in or

register. And to ensure the quality of the discussion, our moderating team will review all comments and may edit them for clarity, length, and

relevance. Comments that are overly promotional, mean-spirited, or off-topic may be deleted per the moderators’ judgment. All postings become

 JOIN THE CONVERSATION

Inventory-Driven Costs https://hbr.org/2005/03/inventory-driven-costs#

12 of 13 10/7/2015 5:29 PM

the property of Harvard Business Publishing.

Inventory-Driven Costs https://hbr.org/2005/03/inventory-driven-costs#

13 of 13 10/7/2015 5:29 PM