Public procurement innovation:
pharmaceutical tendering

DeepDive podcast, available on Chrome

https://notebooklm.google.com/notebook/9eace7f3-fa0c-4e32-9b0f-f3b8a1527106/audio

 

 

BRIEFING NOTE

Design of recurrent procurement auctions

 

Source :

Bokhari, F., Ennis, S., Salmon, T., and Vega, C."The Design of Recurrent Procurement Auctions", CCP Working Paper 24-04, https://cdn.sanity.io/files/hr4v9eo1/production/09dd7e5c3ade1a00f5ac730d6400b6843c76367d.pdf

 

 

Executive Summary

This working paper examines how different ways of setting bid limits, called dynamic bid caps , affect procurement auctions where items are repeatedly bought. The study uses experiments to compare a baseline with no cap to two types of dynamic caps: one based on the previous lowest bid in the same market (SBC) and another based on a different market (MBC). While both SBC and MBC successfully prevent prices from rising when fewer sellers compete, they significantly increase the rate at which auctions fail because potential sellers are priced out. Ultimately, the research highlights a crucial trade-off between achieving lower prices and encountering extensive market failure when using these dynamic bid cap mechanisms.

​

Key words:

Dynamic bid caps

Auction failure rates

Dampened competition

Procurement auctions

Price trade-offs

Dynamic Bid Caps: Prices vs. Market Failure

​

What does this paper do?

This paper explores the effects of dynamic reserve price mechanisms, specifically referred to as dynamic bid caps, on bidding behavior, market entry/exit, auction failure rates, and efficiency in multi-round procurement auctions. Using controlled laboratory experiments, the study compares three scenarios: no bid cap (NBC), a dynamic bid cap based on the lowest bid in the same market from the previous period (SBC), and a dynamic bid cap based on the lowest bid in a different, matched market from the previous period (MBC). The key findings indicate that while both SBC and MBC successfully address the issue of increasing prices observed in NBC auctions due to dampened competition after bidder exit, they lead to significantly higher rates of auction failure as more bidders are priced out and exit the market. This presents a trade-off between lower auction prices and extensive market failure.

 

Main Themes and Key Ideas:

 

The Problem of Dampened Competition in Repeated Auctions: In repeated procurement auctions, less efficient firms tend to exit over time, reducing the number of competitors. This leads to less aggressive bidding and an expected rise in auction prices.

 

"However, when bidding to sell the same object is done repeatedly, inefficient firms tend to exit the market over time, resulting in a dampened level of competition. With fewer sellers competing for the contract, bidding becomes less aggressive, and prices are expected to rise."

 

Dynamic Reserve Prices as a Potential Mitigation: Dynamic reserve price mechanisms, or bid caps, set based on previous auction outcomes, are explored as a way to potentially keep prices low even with fewer sellers.

 

Dynamic Bid Cap Mechanisms Tested:

 

Single Market Dynamic Bid Cap (SBC): The bid cap for the current round is the lowest bid from the previous round in the same market . This mechanism is potentially vulnerable to manipulation and tacit collusion.

 

Multi-Market Dynamic Bid Cap (MBC): The bid cap for the current round is the lowest bid from the previous round in a different, matched market . This is intended to address the endogeneity issue of SBC and restore competition.

 

No Bid Cap (NBC): A baseline scenario with no dynamic reserve price imposed.

 

Bidding Behavior Across Treatments:

 

In the first round, bidding behavior is statistically similar across all three treatments.

 

In subsequent rounds (2 and 3), bidding in NBC auctions becomes significantly less competitive, leading to higher prices.

 

Bidding behavior in SBC and MBC auctions becomes deceptively more competitive in subsequent rounds compared to NBC.

 

"Surprisingly, bidding behavior remains similar across bid cap institutions during the first round. In subsequent rounds, bidding becomes deceptively more competitive in auctions with bid caps, but unexpectedly resulting in destroying markets."

 

"RESULT 1. There is no significant statistical difference across treatments for round 1 bidding. In rounds 2 and 3, we find that bidding behavior in NBC auctions becomes increasingly less competitive relative to both SBC and MBC auctions."

 

Impact on Auction Prices:

 

NBC auctions show an increase in average prices over subsequent rounds, consistent with theory regarding reduced competition.

 

SBC and MBC auctions, while starting with slightly higher average prices in Round 1, result in lower average prices in Rounds 2 and 3 compared to NBC.

 

"RESULT 2. On average, prices are lower with SBC and MBC than in NBC auctions."

 

"In subsequent rounds, auction prices rise in the NBC auctions, while in both the SBC and MBC cases, prices drop."

 

Surprisingly, prices in SBC auctions are even lower than in MBC auctions in Round 3, suggesting more aggressive bidding despite the endogenous nature of the SBC mechanism.

 

Impact on Bidder Entry and Auction Failure:

 

The dynamic bid caps (SBC and MBC) significantly reduce the probability of a bidder entering the auction in subsequent rounds.

 

This reduced entry leads to a much higher likelihood of auction failure in SBC and MBC treatments compared to NBC.

 

"RESULT 3. The auctioneer's problem of high prices is addressed, but bidders are more likely to exit markets with the SBC and MBC than in NBC auctions."

 

"On average, both bid cap institutions make failure 56% more likely than without a bid cap..."

 

In Round 3, the failure rate can reach up to 2 in 3 auctions in SBC and MBC scenarios, compared to an average failure rate of up to 2% in NBC.

 

Most bidder exits in failed auctions occur when the expected surplus from winning is negative.

 

Efficiency Considerations:

 

Efficiency, measured by the frequency with which the lowest-cost bidder wins, is higher in SBC (84.25%) and MBC (80.00%) auctions compared to NBC (77.53%).

 

Another measure of efficiency, the ratio of the winning cost to the lowest cost, shows similar average results across all treatments.

 

Trade-off between Lower Prices and Market Failure: The study highlights a clear trade-off: while dynamic bid caps can effectively lower auction prices, they do so by pricing out bidders, leading to a significant increase in auction failures.

 

"As a general result, we observed a clear trade-off between lower auction prices and extensive auction failure."

 

"With the introduction of a dynamic bid cap, we find that bidders exit markets so much that auctions are almost 60% on average more likely to fail with bid caps."

 

Important Facts:

 

The study uses a controlled laboratory setting with three experimental treatments: No Bid Cap (NBC), Single Market Dynamic Bid Cap (SBC), and Multi-Market Dynamic Bid Cap (MBC).

 

The auction format is a sealed bid, first price auction (low price auction) where the lowest bid wins.

 

Sellers have idiosyncratic costs that vary across sequences and rounds.

 

Participants start with a balance of 50 ECUs (Experimental Currency Units), with 1 ECU = 10 Philippine Pesos.

 

A participation fee of 2 ECUs is charged for each round a bidder chooses to enter.

 

Bankruptcy rules are in place, with bankrupt participants replaced by alternates.

 

The experiment consists of 10 sequences, each with up to 3 auction rounds.

 

In Rounds 2 and 3, bidders can choose whether or not to participate.

 

The bid cap in SBC and MBC is based on the winning bid of the previous round (in the same market for SBC, in a matched market for MBC).

 

Implications:

The findings suggest that implementing dynamic bid caps in repeated procurement auctions, while effective in controlling price increases driven by reduced competition, can have severe unintended consequences in the form of widespread auction failure. For auctioneers, such as governments procuring essential goods, this means that the benefit of lower prices might be offset by the inability to acquire the needed items. The study also notes that if failed auctions are re-run with inflated reserve prices, this could negate the price-reducing effect of the bid cap and potentially lead to even higher prices than in the NBC scenario, in addition to the costs of re-running auctions.

 

Further Research:

The paper notes that this is ongoing work and the results are preliminary. Further analysis is needed to fully understand the long-term effects of these mechanisms and explore potential strategies to mitigate the high failure rates while maintaining price control.

 

 

Paper Summary Initial Draft By NotebookLM