Matthijs Lof

Rational Speculators, Contrarians, and Excess Volatility

The vector autoregressive approach for testing present value models is applied to a heterogeneous-agent asset pricing model using historical observations of the S&P 500 index. Besides fundamentalists, who value assets according to expected dividends, the model features rational and contrarian speculators. Agents choose their strategy based on evolutionary considerations. Supplementing the standard present value model with speculative agents dramatically improves the models ability to replicate observed market dynamics. In particular, the existence of contrarians can explain some of the most volatile episodes including the 1990s bubble, suggesting this was not a rational bubble. Data, as supplemental material, are available at http://dx.doi.org/10.1287/mnsc.2014.1937 . This paper was accepted by Itay Goldstein, finance.

Noncausality and the Commodity Currency Hypothesis

This paper provides new evidence on the role of exchange rates in forecasting commodity prices. Consistent with previous studies, we find that commodity currencies hold out-of-sample predictive power for commodity prices when using standard linear predictive regressions. After we reconsider the evidence using noncausal autoregressions, which provide a better fit to the data and are able to accommodate the effects of nonlinearities and omitted variables, the predictive power of exchange rates disappears.

GMM Estimation with Non‐Causal Instruments Under Rational Expectations

There is hope for the generalized method of moments (GMM). Lanne and Saikkonen (2011) show that the GMM estimator is inconsistent, when the instruments are lags of noncausal variables. This paper argues that this inconsistency depends on distributional assumptions, that do not always hold. In particular under rational expectations, the GMM estimator is found to be consistent. This result is derived in a linear context and illustrated by simulation of a nonlinear asset pricing model.

Asymmetric Information and the Distribution of Trading Volume

We use the Kyle (1985) model to characterize the distribution of trading volume as a function of the proportion of informed trade. If most orders are submitted by uninformed traders and are hence uncorrelated, order imbalance is small compared to observed trading volume, and the distribution of trading volume resembles a Normal distribution. On the other hand, if most orders are correlated because they are submitted by informed traders, the order imbalance is large and needs to be absorbed by market makers. The trading volume distribution is in this case more skewed and has a higher coefficient of variation (ratio of standard deviation to mean). We find that the volume coefficient of variation (VCV) is highly correlated to alternative measures indicative of informed trading in the cross section of stocks. Moreover, the VCV sharply decreases after earnings announcements, which resolve information asymmetries. Finally, the post earnings announcement drift is stronger for stocks with a low VCV prior to the announcement.

Slow Trading and Stock Return Predictability

The state of market returns positively predicts the size premium (or the difference in the return on small and large firms). A simple trading strategy that buys small firms and sells large firms following positive market return states, and switches to selling small firms and buying large firms after negative market states yields large, risk-adjusted monthly profits of 1.8%, 3.0% or 4.3% when rebalanced monthly, weekly or daily. Moreover, this predictability is also present in actively traded ETFs and in recent years. We uncover that trading patterns by institutional investors contribute to the predictability. Specifically, when rebalancing portfolios, institutional investors appear to execute trades in large-cap stocks swiftly, but are slower in trading small firms.