The growing prevalence of stock market chat rooms and social media suggests that communication between traders may affect market outcomes. Using data from a series of laboratory experiments, we study the causal effect of trader communication on market efficiency. We show that communication allows markets to convey private information more effectively. This effect is robust to a wide range of information settings. The presence of insiders limits the impact, whereas posted reputation scores in the communication platform magnify it. These findings illustrate the need to consider social interactions when designing market institutions to leverage the social motives that foster information aggregation. This paper was accepted by Axel Ockenfels, behavioral economics & decision analysis. Supplemental Material: The e-companion and data are available at https://doi.org/10.1287/mnsc.2023.4967 .

We develop a novel experimental paradigm to study the causal impact of trading algorithms on informational efficiency, liquidity, and welfare. In our design, public information about the asset value is revealed during trading, which gives algorithms a reaction speed advantage. We distinguish market-order (aggressive) and limit-order (passive) algorithms, which replace human traders from the baseline markets. Relative to human-only markets, limit-order algorithms can improve welfare, although human traders do not benefit, as the surplus is captured by the algorithms. Market-order algorithms do not significantly change welfare, though they do lower human traders' profits. Both types of algorithms improve price efficiency, lower volatility, and increase the share of profits for unsophisticated human traders. Our results offer unique evidence that non-exploitative algorithms can enhance welfare and be beneficial to unsophisticated traders.

This Journal of Behavioral Finance issue is dedicated to Vernon Lomax Smith on the occasion of his 95th Birthday. In addition to Vernon’s Birthday, it is also the 20th Anniversary of his 2002 Nobel Prize in Economic Sciences. Vernon is the founder of the field of Experimental Economics and Finance, and his work has spanned a variety of areas in Behavioral Finance. Indeed, Vernon was a founding member of the board of editors for the Journal of Behavioral Finance. Vernon authored the seminal papers on market bubbles in the laboratory (e.g., Smith et al. 1988; Porter and Smith 1994, 1995; Caginalp, Porter, and Smith 1998) along with scores of papers on auctions and auction design (e.g., Smith 1966; Cox, Smith, and Walker 1988; McCabe, Rassenti, and Smith 1991). In terms of individual decision-making and risk, Vernon has examined individual rationality versus market rationality (e.g., Knez, Smith, and Williams 1985). He has also provided deep insights on individual play in game theory (e.g. Hoffman, McCabe, and Smith 1996; McCabe, Rassenti, and Smith 1996; Smith and Wilson 2018). Vernon’s work has greatly influenced several generations of economists and will have a lasting place amongst the many tomes dedicated to the understanding of human behavior. The papers published in this special issue can be broadly grouped into two research areas that Vernon has greatly influenced, market behavior and individual behavior. For market behavior, the paper "Informational price cascades and non-aggregation of asymmetric information in experimental asset markets," by Jason Shachat and Anand Srinivasan, shows that asset markets are not good at aggregating asymmetric information concerning a common dividend. The authors find aggregation fails because prices lock into homegrown norms that form informational cascades. This paper also examines the timing at which information is revealed to traders. When information is released to all traders early in the trading period, prices reflect aggregate information better. This paper is in contrast to Vernon’s famous 1962 paper “An experimental study of competitive market behavior,” where he found that consumer spending markets with defined sellers and buyers can find the competitive market equilibrium using the double auction market. This established the foundation that markets can effectively aggregate information, which became the standard result for many years. However, in 2012, Vernon and his colleagues challenged this result for markets with durable goods (Dickhaut et al. 2012). They found that the information aggregation process is disrupted when goods can be resold. Ever the scientist, Vernon continues to challenge results the profession thought were settled. The second paper on asset markets, "Return predictability in laboratory asset markets," by Zhongming Cheng and Shengle Lin uses laboratory experiments to investigate the positive correlation between retail order imbalance and short-term excess returns. Using the famous design from Smith et al. 1988, the authors find that retail order imbalance in period t positively predicts returns in period tþ 1 (this result replicates the findings of numerous studies with the Smith design). Andrade et al. 2016 further shows that return predictability from order imbalance is increased in laboratory markets with "excited" participants. In the paper in this issue, the authors also test their lab finding on the role of investor sentiment in predicting stock returns in the U.S. stock markets. As in the laboratory experiments, order imbalance in U.S. stock markets significantly predicts returns in high sentiment trading sessions. In the classic paper by Cox, Roberson, and Smith (1982), "Theory and behavior of single object auctions," auction theory predicts the outcomes of the first price sealed bid auction and Dutch clock auction should be identical. To test this prediction, they designed and conducted experiments which found

There is an ongoing debate regarding the degree to which a forecaster’s ability to draw correct inferences from market signals is real or illusory. This paper attempts to shed light on the debate by examining how personal characteristics do or do not affect forecaster success. Specifically, we investigate the role of fluid intelligence, manipulativeness, and theory of mind on forecast accuracy in experimental asset markets. We find that intelligence improves forecaster performance when market mispricing is low, manipulativeness improves forecaster performance when mispricing is high, and the degree to which theory of mind skills matter depends on both the level of mispricing and how information is displayed. All three of these results are consistent with hypotheses derived from the previous literature. Additionally, we observe that male forecasters outperform female forecasters after controlling for intelligence, manipulativeness, and theory of mind skills as well as risk aversion. Interestingly, we do not find any evidence that forecaster performance improves with experience across markets or within markets. This paper was accepted by Axel Ockenfels, behavioral economics and decision analysis.

We assess the effect of the cognitive make-up of market participants on the informational efficiency of markets. We put forth that cognitive skills, such as cognitive reflection, are crucial for ensuring the informational efficiency of markets because they endow participants with the ability to infer others' information from prices. Using laboratory experiments, we show that information aggregation is significantly enhanced when (i) all participants possess high levels of cognitive sophistication and (ii) this high level of cognitive sophistication is common information for all participants. Our findings shed light on the cognitive and informational constraints underlying the efficient market hypothesis.

The impact of high-frequency trading (HFT) strategies on market quality has been debated in both public forums and academic studies for years. Although some consider HFT to be inherently bad, others view it as providing important liquidity to markets. Thus, the question regarding HFT’s impact on market quality remains open. To address this, the authors conducted two sets of controlled laboratory experiments, one with and one without an HFT robot trader. The authors focus on two aspects of HFT trading strategies that have the potential to affect market quality negatively: arbitrage and directional trading. Indeed the authors designed the HFT robot to have a perfect view of the market before executing a trade to provide it the best chance to influence the market. The introduction of this HFT robot had a significant positive impact on trading volume and bid depth, but it had a negligible impact on other market quality indicators such as efficiency, price volatility, bid-ask spread, and book depth. Thus, the authors find that in this polar case, HFT is neither a drain on nor a boost to market quality, suggesting that HFT trading, in its worst case, has a benign effect on the market. TOPICS: Futures and forward contracts, portfolio construction, portfolio theory Key Findings • We seek to examine the effect of high-frequency trading on market quality using controlled laboratory experiments. • We find no statistically significant effects for most measures of market quality when we introduce high-frequency trading. The only measures that are affected by high-frequency trading are trading volume and bid depth. • High-frequency traders are effective middlemen and invoke a response from traders in terms of bidding behavior.

The methodology presented provides a quantitative way to characterize investor behavior and price dynamics within a particular asset class and time period. The methodology is applied to a data set consisting of over 250,000 data points of the S&P 100 stocks during 2004-2018. Using a two-way fixed-effects model, we uncover trader motivations including evidence of both under- and overreaction within a unified setting. A nonlinear relationship is found between return and trend suggesting a small, positive trend increases the return, while a larger one tends to decrease it. The shape parameters of the nonlinearity quantify trader motivation to buy into trends or wait for bargains. The methodology allows the testing of any behavioral finance bias or technical analysis concept.

In a seminal work, Plott and Sunder (1988) offer support for the rational expectations hypothesis and report evidence that markets with certain features aggregate dispersed information. However, their results are based on only a few observations and our attempt to replicate the key findings of that study with an appropriately powered experiment largely fails. The resulting post study probability that market performance is better described by rational expectations than the prior information (Walrasian) model under the conditions specified by Plott and Sunder (1988) is very low. As a result of our failure to replicate, we investigate an alternate set of market features that combines aspects of the original experimental design. For these markets we do find robust evidence of information aggregation in support of the rational expectations model. In total, our results indicate that information aggregation in asset markets is fragile and should only be expected in limited circumstances.

Apparently contradictory evidence has accumulated regarding the extent to which financial markets are informationally efficient. Shedding new light on this old debate, we show that differences in the distribution of private information may explain why informational efficiency can vary greatly across markets. We find that markets are informationally efficient when complete information is concentrated in the hands of competing insiders whereas they are less efficient when private information is dispersed across traders. A learning model helps to illustrate why inferring others’ private information from prices takes more time when information is more dispersed.

The ability of markets to aggregate diverse information is a cornerstone of economics and finance, and empirical evidence for such aggregation has been demonstrated in previous laboratory experiments. Most notably Plott and Sunder (1988) find clear support for the rational expectations hypothesis in their Series B and C markets. However, recent studies have called into question the robustness of these findings. In this paper, we report the result of a direct replication of the key information aggregation results presented in Plott and Sunder. We do not find the same strong evidence in support of rational expectations that Plott and Sunder report suggesting information aggregation is a fragile property of markets.

We conduct a series of experiments to examine the effects of the make and take fee structure currently used by equity exchanges in the U.S. We examine the effects of these fees on measures of market quality (efficiency, book depth, and the bid-ask spread). We find spreads to be smaller in the presence of make and take fees, and we note that this fee structure seems to induce buyers (moreso than sellers) to compete for rebates from the exchange leading to higher prices and lower profits. To test whether our results are due to the make and take fee structure or are artefacts of trading fees in general, we performed a second set of experiments in which traders on both sides of a transaction were assessed an identical fee. These identical trading fees do not appear to significantly affect our market quality measures.

We use a novel tax mechanism - 'rejected offer reassessment' (ROR) - in laboratory experiments to discourage seller holdout and facilitate land assembly. Under this mechanism, if a landowner rejects a developer's offer, his taxable property value is reassessed to be equal to the rejected offer, increasing his taxes. We find that, relative to a control treatment, ROR discourages the magnitude of seller holdout (but not its frequency) and increases the rate of successful land assembly by almost 60%. It also increases the gains from trade by 22.1% relative to the control treatment, but the difference is not statistically significant.

The multi-group asset flow model is a nonlinear dynamical system originally developed as a tool for understanding the behavioral foundations of market phenomena such as flash crashes and price bubbles. In this paper we use a modification of this model to analyze the dynamics of a single-asset market in situations when the trading rates of investors (i.e., their desire to exchange stock for cash) are prescribed ahead of time and independent of the state of the market. Under the assumption of fast trading compared to the time-rate of change in the prescribed trading rates we decompose the dynamics of the system to fast and slow components. We use the model to derive a variety of observations regarding the dynamics of price and investors' wealth, and the dependence of these quantities on the prescribed trading rates. In particular, we show that strategies with constant trading rates, which represent the well-known constant-rebalanced portfolio (CRP) strategies, are optimal in the sense that they minimize investment risks. In contrast, we show that investors pursuing non-CRP strategies are at risk of loss of wealth, as a result of the slow system not being integrable in the sense that cyclic trading rates do not always result in periodic price variations.

Using simulations and experiments, we pinpoint two main drivers of trader performance: cognitive reflection and theory of mind. Both dimensions facilitate traders’ learning about asset valuation. Cognitive reflection helps traders use market signals to update their beliefs whereas theory of mind offers traders crucial hints on the quality of those signals. We show these skills to be complementary because traders benefit from understanding the quality of market signals only if they are capable of processing them. Cognitive reflection relates to previous Behavioral Finance research as it is the best predictor of a trader’s ability to avoid commonly-observed behavioral biases.

We use laboratory experiments to test the ability of two self-assessment tax mechanisms to discourage seller holdout and facilitate land assembly. Each mechanism requires a seller to declare a price at which he is willing to sell his property. The incentive to overstate the value is mitigated by using the declared price to assess a property tax. The incentive to understate the value is mitigated by allowing developers to buy the property at the declared price. One tax mechanism uses tax formula that is complex to calculate but incentive compatible to elicit sellers’ true reservation values. The second uses a flat tax rate that is easy to implement but not incentive compatible. We find that sellers overstate their reservation values under both tax mechanisms. Nevertheless, both mechanisms increase the rate of successful land assembly by 67% and the gains from trade by more than 120% relative to a control treatment. Given their equal performance, the flat tax rate seems the best option given the easy of its implementation.

Markets are often viewed as a tool for aggregating disparate private knowledge, a stance supported by past laboratory experiments. However, traders’ acquisition cost of information has typically been ignored. Results from a laboratory experiment involving six treatments varying the cost of acquiring signals of an asset’s value suggest that when information is costly, markets do not succeed in aggregating it. At an individual level, having information improves trading performance, but not enough to offset the cost of obtaining the information. Although males earn more through trading than females, this differential is offset by the greater propensity of males to buy information such that total profit is similar for males and females. Looking at individual skills, we find that higher theory of mind is associated with greater trading profit, greater overall profit, and an increased likelihood of acquiring information while cognitive reflection is associated with greater profit but not a greater propensity to acquire information.

Abstract We present a methodology that is designed to examine the efficiency of a market as well as uncover the motivations of traders. Using a data set of 124 000 daily ETF (exchange traded fund) price observations, the relationship between volume and price efficiency of ETFs is measured by regressing the fractional deviation between the ETF price and net asset value against the volume of the ETF relative to its own mean and standard deviation. A nonlinear relationship is found, providing support for the thesis that for low to moderate levels of volume greater volume leads to greater efficiency. However, efficiency diminishes as one attains very high volume. To determine trader motivations and examine hypothesized behavioral effects, we present a methodology whereby valuation is modeled and used as an independent variable so that much of the noise inherent in changes in valuation is removed. One can thus examine an open‐ended list of behavioral effects. The impact of these effects on price changes can be compared with one another quantitatively. Using higher powers of the trend variable, we find a nonlinear (cubic) relationship for both the ETF and the S&P 100 studies that involve 77 000 and 185 000 data points, respectively.