Aggregated surveys of expected returns exhibit low correlation with one another and limited comovement with valuation ratios. Using a latent variable framework, I decompose the cross-sectional variation into heterogeneous expectations and idiosyncratic measurement error. The results indicate that systematic disagreement among market participants, rather than measurement error, is the dominant source of variation. On average, common latent factors explain approximately 80% of survey variance, with measurement error accounting for the remainder. Further decomposition reveals that factors associated with extrapolation and cyclicality together explain more than 60% of the total variance. Survey responses thus combine the true market expected return with investors’ disagreement relative to the market, leading to biased inferences about the properties of the market expected return. This persistent disagreement, even under common information, supports models in which agents “agree to disagree” and helps reconcile conflicting empirical findings in the literature.

Presentations: EFA Doctoral Tutorial (Paris), IAAE Annual Conference (Turin), Wharton PhD Brownbag Seminar, University of Luxembourg Brownbag Seminar

Wash trading is a manipulative practice involving fake transactions intended to mislead other market participants. In simple laboratory markets for a one-period security we examine the impact of an algorithmic wash trader on market prices. Our experimental treatments vary the amount of available information as well as the strategy pursued by the algorithmic wash trader. We apply the minimally-intelligent agent model adopted for heterogeneous expectations to predict the outcomes in the experiment. The experimental data show that the wash trader distorts the price discovery of the market in the predicted way. Against our predictions the number of subject-initiated transactions do not decrease with wash trading. Based on the evidence, we suggest amendments to the minimally-intelligent agent model.

Presentations: BSE Summer Forum – Workshop on Computational and Experimental Economics, Experimental Finance Conference (Society for Experimental Finance, Bonn)

This chapter surveys the nascent experimental research on the interaction between human and algorithmic (bot) traders in experimental markets. We first discuss studies in which algorithmic traders are in the researcher’s hands. Specifically, the researcher assigns computer agents as traders in the market. We then followed it up by discussing studies in which the researcher allows human traders to decide whether to employ algorithms for trading or to trade by themselves. The paper introduces the types and performances of algorithmic traders that interact with human subjects in the laboratory, including zero-intelligent traders, arbitragers, fundamentalists, adaptive algorithms, and manipulators. We find that whether algorithm traders earn more profit than human traders crucially depends on the asset’s fundamental value process and the market environment. The potential impact of interactions with algorithms on the investor’s psychology is also discussed.