Do Learning Curves Offer Potential New Insights into Economic Theory?

I recently taught a training class on learning curve theory, which is a basic cost estimating concept for production. Engineers working on aircraft manufacturing in the early- and mid-twentieth century noticed that for a specific aircraft, every time the quantity doubled, the production cost dropped by a constant percentage. Since that time, this propensity has been found to apply to many industries, from the production of helicopters and missiles to hand-held calculators and cars. This tendency for cost to decrease as a function of quantity seems an apparent contradiction to basic economics, in which the supply curve exhibits a positive correlation between quantity and cost. See the graph below for a comparison.

For the purposes of comparison, I am taking a liberty by using the term price in relation to learning curves. The y-axis for leaning curves is actually cost. The difference between price and cost is that price is equal to the cost plus profit. However, profit is typically a percentage of cost, so a learning curve with price as the y-axis should have the same shape as one with cost on the y-axis.

Economic theory has largely neglected the empirical phenomenon of learning. I could not find any mention of it in the graduate-level textbooks I looked at. Fortunately, one of my neighbors is a retired economics professor, so I asked him about learning curves in economic theory. He loaned me a textbook titled Managerial Economics. Published in 1976, it devotes a chapter to learning curves and states: “The objective of this chapter is to introduce a very useful tool that has been employed in industry since the 1930s but has as yet received limited attention by academicians.” It later states “Economists have long been aware of the existence of economies of mass production and decreasing costs in some industries. Increasing economies are typically said to occur in the early stages of development of an industry followed by a longer (sometimes much longer) period of constant costs that eventually blend, as the industry matures, into a final phase of rising unit costs.” The author essentially admits that learning curves portray a much different picture than the traditional theory of the firm taught in microeconomics, but considers it to be largely a transitory phenomenon that dissipates in the long run. But as my friend and colleague Doug Howarth has pointed out to me, the learning phenomenon can persist for extended periods of time. For example, the cost of solar photovoltaic modules has persistently seen a 20% reduction in cost every time production has doubled since as far back as the 1970s. Called Swanson’s Law, the graph below illustrates this phenomenon.

One distinction between the learning curve and and the supply curve is that while the two graphs have the same axes – price and quantity – they are not as similar as they appear. There is a difference in the treatment of time between the two. While learning takes place over time supply and demand curves do not include time as a dimension. Instead economic theory keeps it fixed, but this is not made plain in economic theory. This key difference was noted by the late UCLA economist Armen Alchian, who studied learning curves in aircraft production during World War II. I will expound in this distinction in my next blog post.

If you enjoy my work, you should check out my recently published book Solving for Project Risk Management: Understanding the Critical Role of Uncertainty in Project Management. You can read Chapter 1 for free and purchase from Amazon or Barnes and Noble.