Previously: Roots of Progress Fellowship
The US is out of touch with manufacturing, and it is obvious to anyone involved in production. When a reasonably smart person who understands technology walks into a production setting, they often lack a key skill: how to talk about production.
For businesses involved in manufacturing, production is where they have the most control. That control must be wielded rigorously from conception. The details matter, quantification matters, and even the risks that cannot be quantified must be discussed. If the US is to build things again, better, faster, and cheaper than the world has ever seen, we must be a culture steeped in the language of manufacturing: cost accounting.
Cost accounting is not like accounting for your taxes: it is not about the past. Rather, cost accounting is a rigorous model of the future. When you understand the shape of cost accounting, you can ask the questions that let you see around corners to spot dead-ends, frauds, and better tomorrows. The fate of entire companies has pivoted on cost accounting: Intel famously told Steve Jobs they would not build the chip in the iPhone based on a cost model. That fateful decision led to them making essentially no revenue from smartphones ever.
This is not a textbook on cost accounting, but a primer. Cost accounting is one of the pillars of understanding how and why material progress happens. It is easy to pick up and generalizes across manufacturing.
The Shape of Cost Accounting
In Andy Grove’s High Output Management, he explains the essence of manufacturing through the lens of the Breakfast Factory:
To understand the principles of production, imagine that you’re a waiter, which I was while I went to college, and that your task is to serve a breakfast consisting of a three-minute soft-boiled egg, buttered toast, and coffee. Your job is to prepare and deliver the three items simultaneously, each of them fresh and hot. The task here encompasses the basic requirements of production. These are to build and deliver products in response to the demands of the customer at a scheduled delivery time, at an acceptable quality level, and at the lowest possible cost.
Let’s revisit the Breakfast Factory, starting with a brief overview through schedule and quality before we get to cost. If you have ever sat like an idiot without food while everyone else had theirs, you understand the importance of getting the schedule right. However, if you have ever had to cook a large meal for a holiday, you understand the challenge and planning required to execute to a tight schedule.
Quality, as defined by the timeless words of Toyota, means meets spec. It does what it is supposed to do. Quality is not free. Perfect eggs require control of cooking temperature and time. In all forms of manufacturing, tight tolerances, specific shapes, or exotic materials can restrict the suitable kinds of fabrication processes, generally because they have lower throughput.
Conceptually, every manufacturing line can be reduced to two types of costs: direct and indirect. For the math inclined:
Costs = Direct costs + Indirect costs
Each of these kinds of costs can also be split into fixed and variable. Fixed costs you pay regardless of production level, like the lease on the restaurant. Variable costs scale with the level of production, like buying more eggs to serve more customers.
Direct costs can be directly traced to the product. At the Breakfast Factory, direct costs include the people who make the food as well as the eggs, coffee, butter, bread, and the coffee maker and other tools used specifically for making food. The indirect costs cannot be directly traced to the product. At the Breakfast Factory, that could include electricity, natural gas, a refrigerator, air conditioning, electronic security, cleaning materials, and waste disposal. Each of these costs has their own progress curve that can be measured. Let’s step through each kind of direct cost and indirect cost, before pulling back to the high level view of thinking about costs.
The first kind of direct cost is labor, the cost of people involved directly in making the food. Labor rate usually scales with time, because you pay people for their time. Labor rate is the amount paid per hour to the people making the food. To get actual labor costs, multiply by the number of workers and the hours worked:
Labor Costs = Labor rate x number of workers x hours worked / number of good units produced
In this case, we want to know how much each unit costs, so we divide by the number of good units produced. We specify good units because we do not want to be in the business of delivering food that was dropped on the floor. Be careful to include all of the labor associated costs, typically the wage plus any benefits. The wage is typically tied to supply and demand at different skill levels. You could imagine in our Breakfast Factory where the lowest skilled and lowest wage workers watch the machine make soft-boiled eggs while a higher-paid chef makes perfect omelets by hand. Progress frequently takes the shape of helping lower-skilled employees perform tasks of higher complexity. If we decided to expand our offerings from just soft boiled eggs to eggs over easy, a hand-held tool that helps workers flip eggs perfectly every time would make it easier to find staff for that task without requiring extra training. To go high-tech for a moment, some of the early studies of the impact of AI on office workers have reported similar effects.
Material is not much more complex. It is the cost of the eggs, butter, toast, water, and coffee. For each product the cost would look something like:
Material Costs = (Cost of materials used in good units and scrap) / Number of good units
How much we pay could be a function of how picky we are about the quality of each. We might pay more for a supplier we trust to deliver good products on time. The waste itself is important to track. Food can be overcooked or dropped, and food can be sent back if there was a miscommunication about the order. Both of these result in waste. The cost of waste must be factored into the cost of the units you actually sell. Here, progress looks like minimizing waste, reducing the cost of inspections, and having short, responsive supply lines. That is operational progress, which is useful, but the biggest impact often comes from material progress. Imagine a wonder egg that always tastes amazing, always cooks perfectly, and never spoils. How much would you pay for the wonder egg versus a normal egg? The cost model is essential to understanding the impact of bringing that innovation into production.
Machine rate is deceptively complex. Again, the simple version is:
Machine costs = Machine rate x hours used / number of good units produced
The machines themselves are easy to pick out: the boiler, the toaster, the coffee machine, and the pots and pans used to prepare the food. The kinds of machines we buy are tied to the schedule: how much will we need, by when? If everything has to go out together, it doesn’t make sense to spend a lot of money on a toaster that can toast in 10 seconds while you’re still waiting for the eggs to finish. We have to find an efficient solution for throughput and cost, and that is very hard. A huge part of progress in material goods is figuring out how to solve this problem with better technology and processes. So we use benchmarks and assumptions to model the cost of the machine's useful life.
For example, imagine the toaster saw constant usage for 4 hours during breakfast this morning. If it can only last for 4000 hours of use, we have consumed 0.1% of the value of the machine. We can expect to have to buy a new one in 1000 days. In some cases, a machine’s useful life is best matched to a per-unit basis; in others, its useful life is better aligned to the duration of use. The answer lies with whichever better matches to the actual cost incurred.
Machines are classically where progress happens. Progress here could mean all kinds of different optimization points. A high-pressure boiler could cook eggs faster. A bigger toaster might fit ten slices at once (i.e. batch production) more efficiently. A new kind of toaster could toast faster and more consistently. All of these machines could be integrated into one machine that does everything at the touch of a button. Machine innovation is a key path to abundance.
To summarize our direct costs:
Direct costs = Labor cost + material cost + machine cost
Progress is most visible in direct costs. Producing more with less labor at higher quality is essentially the history of the industrial revolution. But even this belies the revolutions in indirect costs, such as adding electricity, refrigeration, and air conditioning to buildings and vehicles. Forklifts and security cameras completely change the productivity of each task. Indeed, these kinds of innovations in indirect costs can enable entirely new kinds of factories to be built. Adding solar and a battery to the Breakfast Factory may get us all the power we need during the day for our few hours of breakfast production (and maybe that pushes us to consider a fully electric boiler instead of a gas boiler).
Total costs = Direct costs + Some part of indirect costs
There is much more that could be said about cost accounting, as it is truly the lingua franca of production. This framework generalizes to every piece of production in the modern age. Building a 50,000 lbs press to fabricate monolithic parts is better than fabricating multiple parts and assembling them (a per-part comparison requiring analysis across several lines). Conversely, cost accounting can tell you it’s better (likely for quality, throughput, and cost) to assemble a rocket out of components fabricated in multiple ways than attempting to fabricate the rocket as a monolithic, 3D printed device.
If the US is to make a future abundant with food, energy, water, housing, pharmaceuticals, it will have to re-industrialize with sound policy and investments. Those decisions will be spoken in the tongue of cost accounting, transcribed in the tomes of spreadsheets, garishly copied into slides.
Thanks to Rob Tracinski and Mary Hui for help with this post.
As a form of cost-benefit analysis. :)
I found the following equation a bit confusing:
Material Costs = Cost per unit x amount of waste per good unit
Could you clarify whether you meant:
1. Material Costs = Cost per unit + Cost of waste per good unit
or
2. "amount of waste per good unit" is the ratio "Total Material" / "Material in Good Units" (which is 1.0 with no waste and 2.0 if half the material is wasted)? This math works but assumes that the good unit is constructed entirely out of waste, which was the part I found confusing.