Seller Concentration and Marketplace Power: When More Sellers Does Not Mean More Competition
June 2026 | Marketplace Economics | Industrial Organization | Empirical Research Note | ← Back to Blog
This analysis draws from a working paper on profit-aware pricing in two-sided marketplaces, examining seller concentration, pricing power, and competitive dynamics across 110,840 transactions from the Olist Brazilian e-commerce marketplace.
The Starting Point
96.3% of products in the marketplace I analyzed had exactly one seller.
In that setting, pricing power is bounded less by direct competitor reaction and more by buyer demand elasticity and outside options. With a single seller and limited substitutes, the seller sets the price and the buyer decides whether to accept it. Competitive discipline comes from the demand curve, not from rival pricing.
The overall seller concentration tells the same story. The Gini coefficient across sellers was 0.75. The top 20% of sellers generated 82% of total revenue. The bottom 50% generated 3.3%. This is an extremely concentrated revenue distribution, consistent with the kind of winner-take-most dynamics observed in other platform markets.
"More sellers does not automatically mean more competition. The market structure and the mechanism behind it matter."
The More Interesting Result
The concentration finding is striking but not surprising. Single-seller dominance in a long-tail e-commerce marketplace is consistent with standard platform economics. What was more interesting appeared among the remaining 3.7% of products with multiple sellers.
Among products with more than one seller, 84% had exactly two sellers. True multi-seller competition was rare. And within that group, an unexpected pattern emerged.
One might expect duopolies to show more price dispersion than markets with three or more sellers, since fewer competitors typically means less pricing pressure. The data suggested otherwise.
Products with two sellers exhibited lower price dispersion than products with three or more sellers. The median price coefficient of variation in duopolies was 0.058. In markets with three or more sellers it was 0.108. Duopolies showed tighter, more coordinated pricing. Markets with more sellers showed greater price spread.
Why Might This Happen?
Two mechanisms are consistent with this pattern.
The first is tacit coordination. One explanation consistent with industrial organization theory is that two sellers can monitor each other's pricing more easily, making tacit coordination more sustainable. When only two sellers are present, each can observe the other's price directly and respond to any deviation. The punishment mechanism is credible because both parties know retaliation is immediate and certain. As additional sellers enter, monitoring becomes more costly and the coordination equilibrium becomes harder to sustain.
The second is product differentiation. With more sellers, firms may compete along different dimensions: price, quality, reputation, shipping speed, or service. This leads to greater price dispersion even as nominal competition increases. Sellers in oligopolistic markets are not necessarily competing head-to-head on price. They may be differentiating across price tiers, which produces dispersion rather than convergence.
Both mechanisms can operate simultaneously. The data cannot cleanly separate them. But the pattern is consistent with both, and the IO intuition behind each is well established.
The Platform Design Implication
Platforms often assume that increasing seller count improves competition and benefits buyers through lower prices. This assumption is intuitive but incomplete.
In this marketplace, the transition from one seller to two does not appear to produce the pricing patterns a platform might expect from increased competition. Instead, prices remain relatively close together, a pattern that is consistent with several mechanisms including tacit coordination and limited differentiation. The transition that may matter, if this pattern generalizes, is moving from two sellers to three or more. That appears to be the threshold where coordination becomes harder to sustain and pricing strategies diverge.
This has practical implications for how platforms recruit sellers, how they structure category management, and how they think about competitive health at the product level. A category with two sellers may look competitive on a surface measure of seller count. The pricing data suggests it may not be.
A Note on Scope
These findings are descriptive and specific to this marketplace. The Olist platform operates as an aggregator connecting sellers to buyers across Brazilian e-commerce channels. The market structure observed here reflects the characteristics of that specific context, including the long-tail product distribution, the geographic dispersion of buyers and sellers, and the platform's role as an intermediary rather than a direct retailer.
Whether the duopoly coordination pattern generalizes beyond this setting is an open empirical question. The IO mechanisms that would produce it are general, but the magnitude and threshold effects are likely context-dependent.
Conclusion
Seller count alone does not tell us how competitive a market is. In this marketplace, 96.3% of products had a single seller, making competitive analysis at the seller level largely irrelevant for most of the product catalog. Among the small share of products with multiple sellers, the data is consistent with mechanisms such as tacit coordination and product differentiation that can produce greater price dispersion as seller counts increase.
Market structure matters. The mechanisms shaping seller behavior matter even more. Sometimes the most interesting question is not how many competitors exist, but how they compete.
This analysis is part of a broader working paper on profit-aware pricing in two-sided marketplaces, examining demand elasticity estimation, profit optimization under cost uncertainty, customer lifetime value modeling, and implementation frameworks across 110,840 transactions from the Olist Brazilian marketplace. The full paper is available on my research page.
