8/12/2019 BCG_The_Experience_Curve__lll_Why_Does_It_Work_Jan_73_tcm80-13907.pdf

1/3

The Experience Curve-ReviewedIII - Why Does It Work ?

Cost of value added declines approximately 20-30 per-cent each time accumulated experience is doubled.

This is an observable phenomenon. Whateverthe reason, it happens. Explanations are rationali-zations.

The whole history of increased productivityand industrialization is based on specialization ofeffort and investment in tools. So is the experiencecurve. It is a measure of the potential effect of spe-cialization and investment.

Learning

Workers learn. If they learn to do a task better,they can do it in less time. This is equivalent toproducing more in the same time. Characteristi-cally, output can increase 10-15 percent each timetotal output is doubled. This is the well-knownlearning curve measure of man hour productivityincrease.

Based on the learning curve, labor costs only

should decline 10-15 percent each time accumula-ted experience doubles.

Specialization

When scale of activity increases so that num-bers of people are involved, then it becomes pos-sible to specialize.

If two people are doing the same thing, itbecomes possible to break the task into two parts.One person does all of one half. The other persondoes all of the other half. Each will therefore do

his respective task twice as often for a given totaloutput.

The learning curve described above predictsthat with twice the experience the labor timeshould be reduced 10-15 percent.

Increase in scale permits such specialization.Consequently, each worker will approach a totalexperience at any point in time which would betwice as much as he could have achieved withoutspecialization. Doing half as much but twice asoften equals the same amount of effort but twicethe experience with the task. Consequently, spe-cialization permits 10-15 percent less time per unitor 10-15 percent more output in a given time.

If the scale doubles simultaneously with totalexperience, then these two effects should occur

simultaneously. Costs decline 10-15 percentbecause of learning plus 10-15 percent because ofspecialization. The sum of 20-30 percent costdecline is alone an approximation of the totalexperience curve effect.

Where growth in output increases at anyconstant rate, then change in scale and change intotal experience can and often do occur in paral-lel.

Investment

By definition, a profitable investment is onewhere money spent now results in a future payoutthat is larger than the original investment. All thereturn on investment comes in more output forthe same total cost, but deferred.

If the cost of money is extremely high, then vir-tually no investment can be justified. If the cost ofmoney were zero, then any investment whichwould recover the investment and somethingmore by eternity could be justified.

N 1 2 8

THE BOSTON CONSULTING GROUP 1

R E P R I N T

8/12/2019 BCG_The_Experience_Curve__lll_Why_Does_It_Work_Jan_73_tcm80-13907.pdf

2/3

N 1 2 8

THE BOSTON CONSULTING GROUP 2

The cost decline in experience curves is a par-tial function of rate of investment. The control onthis element is the cutoff rate on added invest-ment. If the cutoff rate is high, costs declineslowly. If the cutoff rate is low, costs decline

rapidly.

Return on investment does result in cost reduc-tion. Without investment, capacity increase cannotoccur and neither can cost reduction at constantcapacity.

A significant part of the experience curve costreduction is the result of return on investment.

Scale

The experience curve effect is the result in partof increased scale. Yet there is no justification forincreased scale unless there is growth. There is noneed to add capacity at all.

With growth there is constant addition of capa-city. Each added increment of unit capacity

becomes a smaller percent of the total capacityunless size of the increment is increased also. Bothcapacity utilization and scale effect are affected by

growth.

The effect of scale is well known though verydifficult to measure precisely. There is, however, aformula which is known to approximate scaleeffect in the process industries.

Capital cost increases by the six-tenths powerof the increase in capacity.

This exponential change is equivalent to anincrease of 52 percent in capital cost to provide a100 percent increase in capacity. The total capitalcost became 152 percent instead of 100. The total

output became 200 instead of 100. The averagebecame 152/200 = 76 percent of 100 percent. Thatis a very common and typical experience curvecost decline rate.

Average production unit size normally increasesin proportion to rate of total output or even faster.If it does, then capital cost should go down as fastor even faster than in proportion to a 76 percent

experience curve.

Since capital tends to displace labor over time,then this scale effect becomes increasingly impor-tant with growth in volume and experience.

There are limits on scale due to load factorsand logistics provided there is a finite total market.But if the total market grows, then scale can beexpected to grow too.

Scale effect applies to all operations, not justprocess plants. Marketing, accounting and all theoverhead functions have scale effects also.

Scale effect alone is sufficient to approximatethe experience curve effect where growth isconstant and scale grows with volume.

For most products, a 70-80 percent slope is nor-mal, with the steeper slope for those where themaximum value is added and where shared expe-

rience with slower growth areas is least. However,it is probable that few products decline in cost asfast as they could if optimized.

It is a known fact that costs are more certain todecline if it is generally expected that they shouldand will.

It must be remembered that experience curvecosts are not accounting conventions. They arecash flow rates divided by output rates. Accoun-ting data is an approximation of this but generallytends to show lower average costs since assets aredeferred recognition of cash expenditures.

The Experience Curve - Reviewed

R E P R I N T

8/12/2019 BCG_The_Experience_Curve__lll_Why_Does_It_Work_Jan_73_tcm80-13907.pdf

3/3

N 1 2 8

THE BOSTON CONSULTING GROUP 3

This means that cost of capital and return oncapital from value added are both included inexperience curve costs. Trading profits or lossesfrom price levels are excluded from this cost calcu-lation.

Experience curve costs on the above basis areprobably more accurate representations of costthan any accounting convention, since they are

based on cash flow only, not projections, andbecause such costs include the cost of capital.

The reasons for the experience curve effect arenot particularly important. The important fact isthat the experience curve is a universally obser-vable phenomenon. If costs do not go down in apredictable fashion, then, and then only, do theunderlying reasons become important. Analysiswill usually show the reasons to be inadequateinvestment, improper value added definitions oroccasionally just mismanagement.

Summary

The experience curve is the result of the combi-ned effect of learning, specialization, investmentand scale. The effect of each of these is anapproximation, and so the experience curve effectitself is also an approximation.

The combination of these factors should permit

a considerably steeper experience cost curve thanis actually observed. However, some additionaloverhead cost is introduced by the need to coordi-nate and plan these changes.

All elements of cost do not have the same expe-rience base. Also, some cost elements share expe-rience with other products.

Consequently, only new and unique productswith completely new cost elements can be expec-ted to go down the cost experience curve with themaximum slope.

Bruce D. Henderson

The Boston Consulting Group, Inc. 1974

The Experience Curve - Reviewed

R E P R I N T