Whether in large auction markets involving worldwide sales of, say, automobiles or in more specialized forums like the Tsukiji, Tokyo, fish market, auctioneers provide a platform for buyers and sellers to interact.[1] They also play an important role in designing the market and have significant discretion in administering the auctions. While auctioneers are colorful figures in the popular imagination, they are largely absent from the academic literature, where auctions are typically assumed to be run by a passive entity whose incentives are aligned with the seller’s. Thus, the literature does not provide much guidance on how one may think about strategic behavior by an auctioneer.

Auctioneers in large markets are typically long-term players interacting repeatedly with many clients. Their ultimate success and profitability depend on how well they manage the externalities without jeopardizing the trust of the many buyers and sellers attending their auctions. In our recent article, “Auctioneers as Market Makers: Managing Momentum in Chittagong Tea Auctions,” we argue that it is useful to think of auctioneers as market makers: organizers who benefit from making the overall market profitable rather than entities that merely sell items in one-off transactions. Moreover, they must carefully balance their incentives with those of their clients to ensure that buyers and sellers trust them enough to make the auction attractive to both sides.

Deciphering the role of auctioneers through data from large auction markets is difficult, because it’s unclear how much auction outcomes depend on auctioneer strategy and how much on market conditions. Our unique data set, hand-collected from tea auctions in Chittagong, Bangladesh, provides a rich and colorful economic environment to examine market design and the role of the auctioneer. We explore how the auctioneer chooses the strategic variables within an auction market based on his own incentives and those of the other market participants. Given the potential for divergence in objectives, his clients (the producers or sellers of tea) may limit the auctioneer’s flexibility in choosing strategies that affect auction outcomes. We find, however, the greater the flexibility, the higher and more transparent the prices. Moreover, our findings offer interesting and insightful lessons for the regulation of market design by investigating how market makers behave in the absence of regulatory constraints.

Our article describes how auctioneers manage a large auction market and shows that their actions depend on the market’s structure and their incentives to maximize revenues from the entire market. A key feature of many large markets is that auctioneers sequentially auction numerous lots that are imperfect substitutes.[2] In such a setting, buyers have the opportunity to bid on a future lot if they fail to win the current lot. As such, they take outcomes from sales of past lots into account to form beliefs about the potential competition for future lots, and their bids depend on this belief. The sale of each lot provides information about overall demand, and that affects the outcome of future lots within that auction. Such informational externality creates price interdependence. How well an auctioneer manages the price interdependence proves to be an important component in our theoretical and empirical analysis. The informational externality coming from auction prices implies that the optimal reserve prices are higher when the goal is to maximize the expected revenue from the entire auction.

To address the empirical challenges of measuring the impact of an auctioneer on market outcomes, we study an auction market in which some auctioneers own a stake in some of the items up for sale. The auctioneer has more discretion in choosing the selling strategy for teas in which he has a stake. We use this fact to study the tension between the incentives of an auctioneer and those of individual sellers. In our data, one auction house is a pure auctioneer, which only sells tea lots from estates owned by clients.  The other is an integrated auctioneer selling tea lots from estates owned by the auctioneer’s holding company (we refer to these as affiliated lots) as well as clients’ lots. Another nice feature of the data set is that the auctioneers provided us with their private notes on the quality of the tea they sold. These allowed us to control for virtually all differences in the quality and other characteristics of tea and to separate product heterogeneity from strategy. By relating the variation in auction outcomes to the level of discretion the auctioneer has over the selling strategy, while controlling for lot quality, we can cleanly discern the impact of his strategic choices.

Chittagong tea auctions are carefully structured, following processes that have been in place for decades. To create symmetric information about supply, detailed information about each lot of tea is provided to all buyers before the auction. This includes ensuring that the auctioneers and buyers evaluate tea quality from the same sample drawn from each lot. It is mainly the uncertainty about demand that is resolved on the day of the auction. Buyers attend the auction with detailed purchase plans for the entire catalog of teas on sale, and the impact of demand is determined during the actual auction.

In our interviews with participants, we were repeatedly told about the importance of figuring out the momentum in prices for the day. If bidders believe that overall demand is high, they will bid more aggressively to avoid being stranded without enough lots. If bidders believe that the overall demand is low, they will bid less aggressively. We empirically verify such interdependence of prices. This feature would be absent if we considered each auction in isolation, as is done in much of the empirical auction literature.

Auctioneers aggregate information regarding overall demand faster than individual bidders do, and they incorporate this information into reserve prices. Withdrawing a lot that is not receiving a high enough bid may be needed to maintain price momentum. This helps coordinate the market, raise overall revenue, and reduce the likelihood of buyers being left stranded without enough lots at the end. An auctioneer who is charged with setting the tone, therefore, would like to choose a higher reserve than sellers who benefit less from this externality. While such actions increase the price conditional on sale, they may lead to some lots going unsold. Thus, this is potentially costly for individual sellers who do not internalize the positive price externality as much as a market maker does. As it is easier to take costly actions with the affiliated lots, the auctioneers use those more frequently to inform market participants about demand.

Our data largely confirms this story. We find that there is positive price interdependence across lots, and that withdrawal of a lot also positively affects the price of subsequent lots. Our theoretical model suggests that such price interdependence should lead to higher reserve prices for affiliated lots. We find that affiliated lots sell less frequently than tea from smaller non-affiliated sellers. Moreover, consistent with the theoretical predictions, affiliated lots generate a higher price. We also do not find any evidence that postponing the sale of a lot increases the price of that specific lot on a future auction day. Furthermore, on average, lots sold by the integrated auctioneer generate higher revenue than do lots sold by the pure auctioneer, even though the pure auctioneer is a more established auction house. These results suggest that higher reserve prices are chosen for affiliated lots to control the price momentum for the entire auction.

By taking costly actions with respect to lots from estates that trust him and give him more control, the auctioneer communicates information about realized demand to improve coordination among participants, and, presumably, increases his own payoff in the process. Viewing the auctioneer as a market maker, as opposed to an insignificant agent in the market, we note that he creates opportunities for all as opposed to waiting to exploit opportunities for himself.

ENDNOTES

[1] For our purposes, auction houses and auction platforms are also considered to be auctioneers.

[2] We refer to the sequence of auctions run throughout a day as an auction and each individual auction within it as a lot.

This post comes to us from Professor Tanjim Hossain of the University of Toronto, Professor Fahad Khalil of the University of Washington, and Professor Matthew Shum of the California Institute of Technology. It is based on their recent article, “Auctioneers as Market Makers: Managing Momentum in Chittagong Tea Auctions,” available here.