This study examines, by commodity, how the intensity of bidding
competition observed in electronic auctions affects wholesale auction
prices and short-term price volatility in the produce division of Garak
Agricultural and Marine Products Wholesale Ma...
This study examines, by commodity, how the intensity of bidding
competition observed in electronic auctions affects wholesale auction
prices and short-term price volatility in the produce division of Garak
Agricultural and Marine Products Wholesale Market. Although prior
work on agricultural prices has largely emphasized price time series
and supply conditions, relatively little is known about how
competitive conditions formed within the auction process are linked to
both price levels and volatility.
To address this gap, we combine daily auction statistics with
lot-level bid logs to construct an item–date panel dataset for 24 fruit
and vegetable items. The sample covers the period from January 3,
2022 to March 31, 2025, and forms an unbalanced panel by retaining
only item–date observations with realized auctions. Because bidding
competition is not directly observed as a single variable, we compute
six daily bidding-behavior indicators for each item, including the
number of auctions, the number of bidders, total bids, bids per
auction, bids per bidder, and the winning rate. We then apply
principal component analysis to summarize their common variation
into a Composite bidding intensity index, denoted CI.
Price effects are estimated using item-specific dynamic regressions
with the log of the daily auction price as the dependent variable,
including CI and its squared term, inbound quantity, lagged price, and
month and day-of-week dummies. Short-term price volatility is
measured using rolling coefficients of variation over the most recent
K auction days (K = 3, 5, 7, 10, 14). We report main results for CV7
and assess robustness using alternative K.
The estimates show that the linear CI term in the price equation is
generally positive across items, indicating that, conditional on inbound
volume and lagged prices, stronger bidding competition is associated
with higher auction prices. In contrast, the volatility response is
heterogeneous across items. Based on the marginal effect of CI on
volatility evaluated at each item’s mean CI, the 24 items are classified
into 10 volatility-increasing and 14 volatility-decreasing items.
Scenario calculations around the mean CI further show that
equal-sized increases and decreases in CI do not necessarily yield
symmetric volatility responses.
Overall, the findings suggest that a bidding-intensity index
constructed from bid logs provides a useful summary of competitive
conditions and complements conventional supply-side variables in
explaining short-run price dynamics. Limitations include reliance on
daily aggregated data and the inability to fully identify potential
endogeneity between bidding intensity and prices.