Long-run expectations in a learning-to-forecast experiment

Abstract

In this paper, we elicit both short and long-run expectations about the evolution of the price of a financial asset by conducting a Learning-to-Forecast Experiment (LtFE) in which subjects, in each period, forecast the the asset price for each one of the remaining periods. The aim of this paper is twofold: on the one hand, we try to fill the gap in the experimental literature of LtFEs where great effort has been made in investigating short-run expectations, i.e. one step-ahead predictions, while there are no contributions that elicit long-run expectations. On the other hand, we propose an alternative computational approach with respect to the Heuristic Switching Model (HSM), to replicate the main experimental results. The alternative learning algorithm, called Exploration-Exploitation Algorithm (EEA), is based on the idea that agents anchor their expectations around the last market price, rather than on the fundamental value, with a range proportional to the recent past observed price volatility. Both algorithms perform well in describing the dynamics of short-run expectations and the market price. EEA, additionally, provides a fairly good description of long-run expectations. JEL: D03 G12 C91

Learning to forecast, risk aversion, and microstructural aspects of financial stability

“…When agents learn from their past errors, and adopt correction mechanisms to account for past deviations of their forecasts from correct values, market oscillations can be dampened, i.e., fat tails of returns PDFs can be reduced. The rationale of such an evidence is that, apart from the purely theoretical hypothesis of always-perfectly informed agents, an adaptive correction scheme reveals to be very useful from the aggregate point of view, as confirmed by experiments done in related literature (see Colasante et al (2017Colasante et al ( , 2018 and Anufriev et al (2013), among others).…”

“…A broader approach is used by Aliabadi et al (2017), who show how agents' behavior varies according to the combined effect of individual risk attributes and to learning abilities to account for errors done in past predictions. Other recent contributions show the impact of self-correcting behavior on long-run expectations, as in Colasante et al (2018). In Colasante et al (2017) an experiment is shown to report that agents use adaptive expectations instead of rational ones and that this may lead to a form of collective rationality (despite the absence of communication among participants), which consists in a robust divergence of predictions from the fundamental value.…”

Long-run expectations in a learning-to-forecast experiment: a simulation approach

“…Looking at the experimental results in LtFEs, especially in the one step-ahead predictions, subjects predict the next price anchoring their predictions around the last realized price. A similar mechanism holds for the long-run predictions, as shown in Colasante et al [2016]. The alternative algorithm proposed in this paper, called the Exploration-Exploitation Algorithm, is based on the empirical as well as experimental evidence that agents anchor their expectations around the last market price conditioning the range of variability of expectations on the past observed price volatility.…”

“…to submit forecasts narrowly centred in p t−1 . This is particularly true for the one-step-ahead predictions for which subjects receive an immediate feedback Colasante et al [2016].…”

“…We use two different values for these parameters following the results ofColasante et al [2016]. Estimating a panel regression model, they show that subjects assign a higher weight to their previous forecasts when forming their the long-run expectations compared to the weight assigned to past forecast when forming their short-run expectations.…”