We have all noticed, or should have noticed, that manufactured products tend to get cheaper with time. The extreme case is computing. The first personal computer I used professionally in 1978 was the Commodore PET. This beast had an 8-Bit processor, 4 kB of memory (kilobytes not megabytes!), a small monochrome screen, and a cassette tape recorder for data storage. In today’s money it cost about £5000. The machine I’m typing this on is about 500,000 times faster, has 2000 times more memory, is supporting two large colour displays, and has 7.5 TB of hard disk storage (I am an untidy filer). It cost about £1000 a year ago.
Following the experience curve
The falling price of computing is an example of learning or experience curves in practice. Products get cheaper and/or better. This was first described in 1936 by Theodore P Wright in a study of aircraft manufacturing. Wright’s Law says that as the number of aircraft coming out of a factory increases, the workforce learns how to do the job faster and better and the time and cost of manufacture decreases. This is the learning curve. The Boston Consulting Group generalised the idea in the mid-1960s into the experience curve. This says that the direct costs per unit fall as you scale up; because of learning, greater efficiency in manufacturing and economies of scale. People use learning curve and experience curve interchangeably, but with growing automation in manufacturing, the experience curve is the one we need to focus on.
The experience curve takes the form of a power-law:
Cn = C1P-a
C1 = direct cost of first unit of production
Cn = direct cost of nth unit of production
P = cumulative volume of production
a = experience rate (how fast unit costs fall as volume increases)
The shape of the experience curve is familiar. Unit costs are high as production begins and then fall exponentially towards a limit as volume increases.
The experience curve is important for innovators, businesses, and policymakers. A new technology is likely to be expensive when first introduced to the market. Does it bring enough benefits to sustain that initial high price, at least amongst early adopters? How can you support the early market until you move far enough down the experience curve for it to be self-sustaining? A socially or environmentally superior technology may be too expensive when first introduced. Should government policy be used to support it through that initial phase? And if so, for how long?
Although we see the experience curve everywhere, we make three mistakes thinking about the difference it might make.
We can’t ignore the experience curve
The first mistake is to be unaware of the experience curve, or to pretend that it is not important. The UK government recently announced a 10-point plan for a green industrial revolution. This included ending the sales of internal combustion engine cars and vans by 2030, a dramatic increase in offshore wind power, making our built environment more energy efficient, and many other initiatives. The fundamental pushback from opponents is always the same – cost. “It is too expensive” or “it will devastate our economy”.
The argument on cost has not been true since the Stern Review in 2006 showed that the benefits of strong, early action on climate change far outweigh the costs of not acting. Since 2006, technology has continued to improve, and costs have continued to fall.
Opponents of renewable energy have not noticed (or choose to ignore) that UK offshore wind farms will provide electricity at lower prices than existing gas plants by 2023. Or that onshore wind and solar are now the cheapest forms of new-build energy supply for two-thirds of the global population. This chart from Michael Liebreich shows the falling cost of PV from 1976 to 2018. During that time costs have fallen from about $80 per watt to $0.25 per watt, and prices will continue to fall.
We see the same pushback on electric vehicles. Yes, they are currently more expensive than internal combustion engine vehicles, but Deloitte expect EV’s to reach parity on total cost of ownership with equivalent ICE vehicles sometime around 2022. “Okay” say the opponents “but it still won’t work because there is no charging infrastructure”. No, we have not built out the infrastructure has yet, but we will. There is no fundamental limit that would be a deal-breaker. The workings of the experience curve guarantee that EV’s and charging points will get cheaper and better.
If your only argument against a green industrial revolution is cost, you have already lost.
The bottom of the experience curve is not zero cost
The second error is to believe that there is no end to the experience curve, and that costs can ultimately reach zero. We have been here before. After WWII and the emergence of civil nuclear power, someone said it would provide energy “too cheap to meter”. That turned out not to be true. Yet you hear the same claim about almost every new energy technology that emerged. Enthusiasts still tell me that fusion power will be too cheap to meter once it gets going.
Union Carbide advert from 1957 promoting nuclear power
That power-law curve does not go on forever. It is sitting on a baseline of fundamental costs that the industry cannot change. Things like raw materials. You can see this in commodity chemicals. I told the story recently of the customer wanting us to become more efficient and cut the price of an ingredient by 30%. Unfortunately, over 80% of our cost was a globally traded commodity raw material. Yes, we could reduce our production costs and be better at raw material buying, and of course we worked very hard on that. But there was no way any amount of learning and experience would significantly reduce our cost per tonne.
Experience changes the market
The final error, often made by enthusiastic entrepreneurs and business unit managers, is to assume that you can use the experience curve to drive up your profitability to stratospheric heights. You can cut costs by increasing volume, but to increase volume significantly you need to sell to different markets. Those additional market segments will often be at a lower price. For most products, there is a price/volume curve and a cost/volume curve, and you can’t hold your margin up against economic gravity forever.
In a previous job, I had an interesting conversation with a business unit manager. We had developed a method of producing some special high-performance surfactants at a much lower price. Because of the price, these were mostly used in high value, low volume, specialist applications. My colleague painted a rosy future where we could take the market from the existing provider by dropping the price just a little below theirs, but because our cost was lower, we would increase the profit. He also planned to bring this high-performance surfactant into adjacent markets at the same high price. Since this niche market was quite important to our competitor, we had to point out to him we risked triggering a price war. Also, the applications he hoped to expand into were not currently using the surfactant because of the high price. In fact, you could plot all the commercial surfactants by price per kilo and volume sold, and they fell on a beautiful power law curve. The bigger the volume sold, the lower the price per kilo. His great opportunity was not as good as it looked.
Treat the experience curve with respect
Experience curves are real, and you should be thinking about them whenever you are introducing a new product or technology to the market.
- Don’t talk your way out of product development if the projected cost is too high. There may be ways to support it until the experience curve kicks in. Perhaps finding a niche application or group of customers for whom the price is worthwhile.
- Separate out the costs you can influence and the cost you can’t. For most products, there is a baseline below which the experience curve cannot take you.
- Increasing volume to cut costs can temporarily increase profitability, but increased volume leads eventually to lower prices.