Gold and silver manipulation: What can be empirically verified?

•Gold and silver price manipulation and price suppression is examined.•A mixture of normal approach decomposes returns into abnormal and control samples.•The proportion of negative runs in the abnormal cluster is greater than in the control cluster.•The average return for negative (positive) runs is lower (higher) in the abnormal cluster than in the control cluster.•The high volatility associated with the abnormal cluster is the driver of the results and not manipulation.

The issue of gold and silver price manipulation, in particular price suppression, is examined. We use a mixture of normal approach to decompose the returns into abnormal and control samples. Price suppression is a form of market manipulation of the runs type, where longer negative runs with lower returns than expected would be observed. To explore whether this form of manipulation can be empirically detected the length of runs and the total return observed during a run were computed for modelled abnormal and control clusters in gold and silver. In both metals the proportion of negative runs in the abnormal cluster is greater than the proportion of negative runs in the control cluster. In both cases the average return for negative runs is significantly lower in the abnormal cluster than in the control cluster. When average returns over positive runs are compared the abnormal group has significantly higher expected returns than the control group. Given the short maximum run lengths in the abnormal cluster and the fact that positive runs have significantly higher average returns in the abnormal cluster than in the control cluster, it is likely that that the high volatility associated with the abnormal cluster is the driver of the results presented in this study, as opposed to manipulation.