COMPETITIVE MARKET BEHAVIOR: CONVERGENCE AND ASYMMETRY IN THE EXPERIMENTAL DOUBLE AUCTION

Barbara Ikica, Simon Jantschgi, Heinrich H. Nax, Diego G Nuñez Duran, Bary PradelskiInternational Economic Review 2023 · DOI

TL;DR

This paper investigates how quickly and reliably markets converge to a competitive equilibrium using experimental double auctions. They found that convergence happens surprisingly fast, within a few trading rounds, even with private information. However, initial prices tend to be biased towards buyers due to their greater initial aggressiveness, a phenomenon that persists across various experimental setups.

What Problem Does This Paper Address?

The central question in economics is whether competitive equilibrium (CE) is a good predictor of real-world market outcomes. Specifically, in double auctions (DAs), where buyers and sellers simultaneously submit bids and asks, it's been debated whether CE can be achieved, especially when traders only have private information about their own valuations. While early experiments showed convergence, there's been ongoing research to understand what drives this convergence: is it the market's central clearing mechanism, or the information disseminated about bids and asks? This paper aims to disentangle these factors by running a large-scale experiment with varying levels of information feedback.

Methods

The researchers conducted 104 controlled continuous double auction experiments involving over 1,700 subjects. They used a custom-built online trading platform called "SciOn." The core of their methodology involved:

  • Varying Information Feedback: They implemented different "treatments" where subjects had varying levels of access to the order book (e.g., "Black Box" with no order-book access, "Open Book" with partial or full access).
  • Measuring Convergence: They used metrics like "Distance to CE price" (DCE) and "Realized Allocative Efficiency" (AEff) to track how close market prices and outcomes were to the theoretical CE over time.
  • Analyzing Individual Behavior: They examined initial bids and asks ("relative demands") and subsequent adjustments within and between trading rounds to understand the micro-level drivers of market behavior.
  • Statistical Analysis: They employed Wilcoxon signed-rank tests and Mann-Whitney U tests to rigorously analyze their findings and test hypotheses about convergence and behavioral asymmetries.

Datasets

The study utilized data from 104 experimental double auction markets, involving 1,751 subjects and generating over 86,000 observations of bids and asks. The data was collected through the "SciOn" online platform and is made publicly available on the Open Science Framework (OSF).

Key Insights and Ideas

  • Centralized Clearing is Key, Not Just Information: The most striking finding is that convergence to CE occurs across all treatments, including the "Black Box" treatment with minimal information feedback. This suggests that the centralized clearing mechanism of the double auction is the primary driver of convergence, rather than the availability of detailed order-book information.
  • Initial Buyer Aggressiveness Drives Downward Bias: Despite convergence, initial prices consistently fall below CE levels. This asymmetry is attributed to buyers being significantly more aggressive in their initial bids than sellers are in their asks. This initial aggressive stance by buyers sets the stage for prices to start low and then gradually rise towards equilibrium.
  • Convergence is Fast and Robust: Convergence to CE happens within a handful of trading rounds (typically after the first five), and this pattern holds across a wide variety of market structures and information treatments. This suggests that the double auction is a remarkably efficient mechanism for price discovery.
  • Behavioral Asymmetries Persist: While convergence is robust, the underlying behavioral differences between buyers and sellers (buyers being more aggressive initially) are also persistent and appear to be inherent to the trading process, not solely driven by feedback mechanisms.

Key Results

  • Convergence within 5 Rounds: Markets converged to CE prices within approximately five trading rounds in most treatments.
  • Initial Prices Below CE: Per-session mean prices were significantly below CE prices in the first four trading rounds, confirming a downward asymmetry.
  • Allocative Efficiency Increases: Realized allocative efficiency (AEff) significantly surpassed 80% by the third round and reached over 90% in later rounds.
  • Buyer Aggressiveness: Median initial relative demand for buyers was significantly higher (around 0.68-0.71) compared to sellers (around 0.19) across treatments.
  • Feedback Effects: While centralized clearing is primary, feedback does influence dynamics. Open Book treatments showed quicker demand reductions than Black Box, suggesting feedback helps accelerate convergence.

Limitations and Future Work

  • Focus on "Explanations": The authors acknowledge that their study focuses on identifying empirical ingredients that explain the observed asymmetries (like buyer aggressiveness) rather than delving into the deeper psychological mechanisms driving them.
  • Generalizability: While the study used a large number of subjects and markets, further research is needed to understand convergence patterns in smaller/larger markets, with different subject pools, and under different DA institutions.
  • Individual Learning Models: The findings open avenues for developing and testing individual learning models to better capture the observed buyer-seller dynamics.
  • Information Design: Understanding how different levels of order-book feedback influence behavior can inform optimal information design in DAs, especially with the rise of automated trading.

Summary updated

Abstract

Abstract We conducted a large number of controlled continuous double auction experiments to reproduce and stress‐test the phenomenon of convergence to competitive equilibrium under private information with decentralized trading feedback. Our main finding is that across a total of 104 markets (involving over 1,700 subjects), convergence occurs after a handful of trading periods. Initially, however, there is an inherent asymmetry that favors buyers, typically resulting in prices below equilibrium levels. Analysis of over 80,000 observations of individual bids and asks helps identify empirical ingredients contributing to the observed phenomena including higher levels of aggressiveness initially among buyers than sellers.
BibTeX
@article{ikica2023competitive,
  author = {Barbara Ikica and Simon Jantschgi and Heinrich H. Nax and Diego G Nuñez Duran and Bary Pradelski},
  title = {{COMPETITIVE MARKET BEHAVIOR: CONVERGENCE AND ASYMMETRY IN THE EXPERIMENTAL DOUBLE AUCTION}},
  journal = {International Economic Review},
  year = {2023},
  doi = {10.1111/iere.12630},
}
Figure