Understanding the Cost Curve Diagram
Most people look at the standard economies of scale diagram and think it's straightforward. It isn't. You've got your U-shaped LRAC curve, the downward slope, the minimum efficient scale point, maybe some shading to indicate increasing or decreasing returns. It looks simple on paper. I've spent years watching people misread this diagram in boardrooms and in exams, so let me walk through what it actually shows and where it falls apart. The diagram plots long-run average cost against output volume. The downward-sloping portion on the left represents economies of scale. Each additional unit costs less to produce because fixed costs spread out, specialization kicks in, bulk purchasing discounts apply, and operational efficiency improves. The flat bottom is the range of constant returns to scale. The upward slope on the right is diseconomes of scale, where coordination problems, bureaucracy, and communication breakdowns start pushing costs back up. What the diagram doesn't show is that the curve shape varies wildly depending on your industry. A software company's LRAC looks nothing like a manufacturing plant's. Software has near-zero marginal cost after the initial build, so the curve stays flat for an enormous range. A steel mill hits its MES much faster and may start experiencing diseconomies well before you'd expect. When someone hands you a generic diagram, treat it as a template, not a blueprint.
I ran into this exact problem a few years ago when a client was trying to justify a facility expansion based on a textbook curve. The diagram suggested they'd hit their minimum efficient scale at around 50,000 units per month and then cruise flat for a while. Reality hit differently. Their supply chain couldn't scale linearly. Vendor lead times stretched, inventory carrying costs spiked, and they actually moved into diseconomies of scale right around 35,000 units. The diagram was technically correct for a frictionless model. The real world has friction. We ended up using a piecewise cost model that broke the operation into three phases instead of one smooth curve. It took longer to build but predicted their actual costs within 4 percent instead of 25 percent off. Here's something most guides won't tell you: the minimum efficient scale point isn't a single number. It's a range. Look at the flat bottom of the U and measure its width. That horizontal span matters more than the lowest point itself. A narrow flat bottom means your cost advantage from scaling is fragile. A wide one means you can absorb demand swings without your unit cost spiking. I've seen companies optimize for the absolute minimum on the curve and then get crushed when demand dipped slightly, pushing them onto the steeper part of the left flank. Another thing people miss is the difference between internal and external economies of scale. The diagram you're looking at usually shows internal economies, the ones the firm controls. But external economies, things like industry cluster effects, shared infrastructure, and a trained labor pool, shift the entire curve downward without the firm doing anything. When you're analyzing a location decision or a market entry, ignoring external economies makes your cost projections artificially pessimistic. Conversely, if you're in a declining industry cluster, external diseconomies can shift the curve up even if your own operations are perfectly efficient.
The diagram also assumes you can adjust all inputs freely in the long run. That's the whole point of the LRAC being long-run. But in practice, "long run" means different things for different businesses. For a SaaS company, the long run might be six months. For a semiconductor fab, it's seven years. When someone says "we're operating below our optimal scale," ask what time horizon they're using. The answer changes whether the statement is useful or just optimistic wishful thinking. If you need a reference diagram, the standard one is easy to find in any intermediate microeconomics textbook or on sites like Khan Academy and Investopedia. But don't stop there. The real skill is knowing which parts of the diagram to trust and which parts to ignore based on your specific context. A generic curve is a conversation starter, not a decision tool.
Get the Full Details
