Can repeated online produce auctions weaken competition?

An auction’s highest bid does not necessarily end the story. In the agricultural markets studied by Dipankar Das, Kingshuk Sarkar and Vivek Jadhav, a seller can reject an offer and try to sell the produce in a later round. Buyers, meanwhile, may encounter the same competitors and alternative lots repeatedly.

The researchers ask how those features interact with information about supply in India’s electronic National Agriculture Market, or e-NAM. Their preprint combines an economic model with auction data from Bishnupur and Bagda in West Bengal. It raises a concern about weak competition, but its observations cannot establish that traders colluded or measure a reliable loss from collusion.

Auctions linked over time

The paper describes a sequence of ascending-price auctions in which multiple lots may be offered across markets. A bidder who wins one lot may need less produce later; a seller who rejects a bid may bring supply back to a subsequent auction. Both decisions can therefore affect later prices and participation.

The model represents buyers with different private valuations and limited demand. It also gives sellers a choice between accepting the current high bid and waiting for an expected future offer. In one simplified comparison, a seller can re-offer a lot without holding costs or deterioration over a short interval. That assumption matters for vegetables: storage costs and spoilage could make waiting less attractive.

Although the paper’s abstract describes full seasonal supply transparency, its detailed model assumes something narrower: bidders know a distribution of possible seasonal supply, while the realized arrival schedule and remaining inventory are uncertain. They update their expectations using observed offers and sales.

The authors use this framework to discuss code bidding: bids that may signal how traders intend to divide lots. For example, two bidders could submit offers that leave each winning a different lot. But a split allocation alone is not proof of coordination; different valuations or needs could produce the same outcome. The paper also states that it has not established the existence of an equilibrium for its full dynamic model under its stated assumptions.

What the market data show

The researchers examine bid summaries for eight vegetables and price, arrival and trading data from the two markets. Participation in the reported bid samples was thin: the summary table shows between one and seven bids per lot, depending on the commodity and sample. That is relevant to the authors’ concern because repeated encounters among a small number of traders can make coordination more conceivable. A small number of bids, however, does not reveal the traders’ intentions.

One diagnostic compares distributions of average bids with those of second-highest bids. The paper reports a statistical relationship called second-order stochastic dominance for seven of the eight commodities, but not for pointed gourd. The authors sometimes describe this as strong support for code bidding; elsewhere they acknowledge that the relationship provides no independent evidence of collusion and can arise mechanically in thin markets. The paper also says detailed historical bid paths and losing bids are not publicly available, limiting what the observed summaries can reveal about signalling or repeated strategies.

For a separate supply analysis, the authors use 47 daily observations for each of five vegetables in Bishnupur, from late June to mid-August 2024. They construct measures of potential available stock from arrivals and traded quantities, using assumptions about short periods of storage. Their regressions report negative associations between those constructed supply measures and changes in the most frequently occurring, or modal, price. These are observational associations: they do not show that sellers rejected bids to raise prices, or that supply changes caused the price movements.

Another set of regressions finds that minimum and maximum prices predict modal prices better than arrivals and traded volumes in the specified models. Because these are related price summaries, that result does not establish that traders ignored supply information or coordinated their bids.

A design question, not a verdict

The model illustrates why a seller’s option to wait could shift a sale between rounds and why auction rules deserve attention when participation is limited. The authors propose giving sellers access to online and offline alternatives while restricting buyers to one market, but acknowledge that the proposal’s effects on welfare and competition need further modelling.

The paper is a preprint submitted to a journal, not an established finding that e-NAM traders colluded. More complete bid histories, evidence of seller decisions and a credible comparison with competitive outcomes would be needed to distinguish strategic coordination from other explanations and to assess its consequences.

The supplied source metadata identifies the version as arXiv:2611.10068v1. The extracted paper confirms that identifier, its authors and the date 6 November 2026; the supplied metadata’s precise publication time and landing-page details have not been independently verified. The metadata uses “eNAM” in its title and abstract, while the extracted paper uses “e-NAM.”