Price Dynamics and Bubble Formation in Learning-to-Forecast and -Optimize Experiments
Tomasz Makarewicz (University of Amsterdam)

In contrast to the classical Rational Expectations hypothesis (RE), price bubbles are prevalent in asset pricing experiments. However, the robustness of this nding has sometimes been attributed to improper experimental design. A typical asset pricing experiment (e.g. Smith et al., 1988) has a declining fundamental price, which may mislead the subjects (Kirchler et al., 2012). On the other hand, Learning-to-Forecast (LtF) experiments Hommes et al., 2005) have a constant fundamental price, but focus on price forecasting, not trading.The goal of our paper is to test the unstable behavior of the positive feedback type of markets in a realistic environment with both trading and stable fundamental price. We study the price dynamics of an experimental asset market with a price adjustment rule and myopic mean-variance traders. We design three treatments where subjects (1) submit a forecast, (2) choose quantity to buy or sell and (3) perform both tasks (mixed). In all treatments, the price does not converge to the RE equilibrium, and shows persistent over- and under-valuation. We fi nd bubbles appear with a higher frequency and magnitude in treatments (2) and (3) compared to treatment (1). This is due to the fact that in the treatments with trading, the subjects chase trend in the price to a higher degree, and act in a less coordinated manner. This shows that a bubble is likely to happen when there is persistent heterogeneity in agents’ decisions and strategies. Moreover, treatment (3) reveals that subjects do not have to behave optimally conditional on their price expectations. Compared to former studies (Hommes et al., 2005), this suggests that the LtF experiment has the design most conducive for individual rationality.

************

Unemployment benefits and endogenous Beveridge cycles
Florian Sniekers (University of Amsterdam)

This paper aims to explain the shape and size of the fluctuations in unemployment and vacancies. Adding Keynesian demand externalities to an otherwise standard search and matching model eliminates the need for exogenous shocks in explaining these fluctuations. Under plausible parameters, the dynamics include a stable limit cycle that resembles the observed counterclockwise cycles around the Beveridge curve. Quantitatively, these endogenous ‘Beveridge cycles’ can explain the volatility and persistence of unemployment remarkably well without any exogenous forces, avoiding the amplification and propagation problems of the standard model. Moreover, the coordination problem resulting from the externalities can be mitigated by increasing the outside option of workers. A higher unemployment benefit may therefore steer the economy to a lower unemployment rate and increase welfare.