I have been reading a book called The Haystack Syndrome by Eliyahu Goldratt. It is mainly about how to design an information system, but it does provide a good summary of the ideas behind the Theory of Constraints (TOC).

He outlines how we use cost accounting to run our businesses, then goes on to use solid cause and effect logic to explain the pitfalls of using these methods. His analysis is very interesting.

Here is a breakdown on his rendition of the history of cost accounting:

Cost accounting was developed at the beginning of the century as a tool to allow throughput and costs to be measured on the same basis. We rightly measure the performance of a company based on its net profit and return on investment (the assumption here is that the Goal of a company is to make more money now and in the future). Net profit is calculated by subtracting the operating expense from the throughput. Here lies the fishhook.

Throughput - what we sell is measured by the subtracting the materials cost from the sales price. Since we sell many different products, each made up of different raw materials, we must calculate this as T = Σ_pThroughput_p

Now what about labour and other overheads? These must be calculated at the company level, thus OE = ΣOE_c.

Thus the net profit for the company is calculated as follows:
NP = Σ_pThroughput_p - ΣOE_c.
We see that we need to use two different bases to make this calculation. This does not invalidate the calculation in any way. The maths is sound. So what's the problem?

The problem is that we cannot ask the 'what if' questions of our company such as "which products are most profitable?". The maths does not lead us to make such measurements.

This is where cost accounting comes into play. It uses a mathematical approximation to put the operating expense on the same basis as throughput. Now the maths becomes much more amenable to asking the 'what if' questions. The approximation is achieved by working out the amount of labour spent on producing each item and then multiplying the total cost by a pre-calculated factor to take into account other non-direct costs such as managers, accountants, building etc. Sounds fair doesn't it.

The maths is now: NP = Σ_p(T - OE)_p

The approximation of cost accounting once was a fair approximation, and relied on the implicit assumption that most of the costs are attributed to the products directly (T>>OE).

However, these days, the distribution of costs has changed. This is best explained in the service sector where OE>>T. Most of the revenue and costs do not come from directly making and selling things, but rather providing a service.

Now the assumption that cost accounting was built on is invalid. We need to re-examine the accounting foundations we use to make local measurements.

One thing for sure that does stay constant is that NP = T - OE.

What is the new system?