Savvy Business
Notes / 03

I think cost accounting caused the problem.

Particularly the habit of turning factory costs into hourly rates, attaching those rates to products and resources, and then using the numbers to manage production.

Previous — making every part more efficient

This is slightly uncomfortable to say

I think cost accounting caused the problem.

That is slightly uncomfortable for me to write. Accounting has been part of my working life for a long time. So has installing software in factories. So has standing on a floor watching the numbers and the work refuse to agree.

I am not saying accounting is useless. We need to know what we spent, what we earned, what we own, what we owe, and whether the business made money.

I am saying something more specific. Taking the costs of a factory, turning them into hourly rates, attaching those rates to products and resources, and then using those numbers to manage production sent us toward local efficiency in the first place.

Once you accept that production is a dependent system, I do not think cost accounting is a useful method for managing it.

An idle hour looks like R1,000

Suppose a machine costs us R1,000 an hour. The exact figure does not matter. Once we have calculated that number, an idle hour looks like R1,000 being wasted.

The machine still depreciates. The operator is still being paid. The factory still has overheads. So the perfectly reasonable management response is: keep the machine running.

Given the measurement, the behaviour is perfectly reasonable.

I am not accusing factory managers of being foolish. I am questioning the measurement that makes this behaviour rational.

Fig. 4 — from an hourly rate to a pile of work
  1. 01

    Factory costs

    Depreciation, wages, electricity, overheads.

  2. 02

    Hourly resource rate

    Those costs become R1,000 an hour.

  3. 03

    Idle time looks expensive

    An unused hour now appears as money wasted.

  4. 04

    Increase utilisation

    Keep the machine running. Spread the cost.

  5. 05

    Produce more units

    Calculated cost per unit falls. The report improves.

  6. 06

    Local efficiency wins

    The factory still ships what it could ship before.

Nothing irrational about this

Factory costs become an hourly resource rate. Idle time looks expensive. We increase utilisation. We produce more units. We spread the costs over more units. Calculated unit cost falls. Apparent efficiency improves.

If we look at that resource on its own, it makes perfect sense.

But the resource is not operating on its own. It is part of a dependent system.

Stop. Keep going.

Back to the three stations. Cutting: 100 a day. Printing: 60. Finishing: 80.

Suppose Cutting has already produced everything Printing can process. From a flow perspective, Cutting should stop. Making another 40 does not increase finished output. It simply creates 40 more pieces of work waiting in front of Printing.

But if Cutting has an hourly rate attached to it, idle Cutting looks expensive. So cost accounting tells us to keep it running.

Flow says stop. Cost accounting says keep going. Those are opposite instructions for the same machine.

Fig. 5 — two instructions, same machine

The value of an hour depends on where it is

Cost accounting asks: what did this hour cost us? Dependency asks: what did this hour change?

An additional hour at Cutting might change nothing about total factory output. An additional productive hour at Printing might allow more finished work to leave the entire factory.

Same unit of time. Completely different consequence.

Cost accounting gives the hour a cost. Dependency gives the hour a consequence.

Fig. 6 — two questions, not the same answer

One philosophy has to win

If we tell a resource that is not setting the pace: only produce what the flow of the system requires — while simultaneously telling it: maximise utilisation so we recover your hourly cost — we are giving the same resource opposite instructions.

Eventually one philosophy has to win.

That is why I do not think cost accounting should be used to manage production in a dependent system.

Accounting is not the same as cost accounting

Businesses still need accounting. We absolutely need to know actual expenditure, revenue, cash, assets, liabilities and profit.

My objection is to allocating those costs down to products, resources and hours, and then treating the resulting calculated costs as though they tell us how production should behave.

Accounting tells us what happened to the money. Cost accounting tries to tell us what each product, resource and hour "cost". I do not think those allocated numbers should be allowed to run the factory.

Goldratt was less polite

Eliyahu Goldratt was considerably less polite about this than I am. In 1983 he presented a paper to APICS titled Cost Accounting: The Number One Enemy of Productivity.

The longer I have worked around factories, the more I understand why.

The measurement made the behaviour rational

We wanted to understand what things cost. So we allocated factory costs to hours. That made unused hours look expensive. So we tried to eliminate unused capacity. That made utilisation important. Utilisation made local efficiency important. And local efficiency encouraged resources to produce whether or not the dependent system needed their output.

The managers were not behaving irrationally. The measurement made the behaviour rational. The measurement ignored dependency.

Cost accounting makes perfect sense until you put the dependency back into the picture.

So let's get rid of the spare capacity

There is an obvious reply. Fine. The problem is that Cutting can make 100, Printing 60 and Finishing 80. The factory is badly balanced. Why don't we simply make every department capable of producing the same amount?

Cutting 100. Printing 100. Finishing 100. Packing 100. No excess capacity. No expensive idle machines. Every resource perfectly matched to demand.

Perhaps we have found the answer. Let's build the perfectly balanced factory. No excess capacity. No idle machines. Nothing wasted. It should be the most efficient factory we have ever built.

Then we'll open the doors on Monday morning.

Next

Let's build the perfectly balanced factory.

Previous — making every part more efficient
A question

If something here is still unresolved, send it.

Later notes will come from the questions people actually ask.

Ask a question