On the directional accuracy of survey forecasts: the case of gold and silver

Abstract

We use a nonparametric market-timing test to study the directional accuracy of survey forecasts of the prices of gold and silver. We find that forecasters have market-timing ability with respect to the direction of change of the price of silver at various forecast horizons. In contrast, forecasters have no market-timing ability with respect to the direction of change in the gold price. Combining forecasts of both metal prices to set up a multivariate market-timing test yields no evidence of joint predictability of the directions of change of the prices of gold and silver.

Keywords:

• gold price
• silver price
• forecasting
• market timing

JEL Classification:

• F30
• D84

Acknowledgements

We thank the foundation ‘Geld and Währung’ from the Deutsche Bundesbank (S126/10081/11) for financial support.

Notes

¹ For recent empirical evidence on the safe haven properties of gold, see Baur and Lucey (2010).

² In earlier research, Schnader and Stekler (1990) and Stekler (1994) apply the market-timing tests of Henriksson and Merton (1981) to study the directional accuracy of real GDP and inflation forecasts.

³ Forecasts for 2003 are not available. Our sample contains data for September (December) 2005/2006, which starts with forecasts for December (March). The corresponding data for 2004, in contrast, starts with one month-ahead forecasts. Because it is not clear whether forecasters indeed kept track of the change in the forecast horizon, we classify the forecasts in analogy to the forecasts made in 2004. Our results are not sensitive to this classification (or to deleting the data altogether). Our empirical results are computed using the free software R (R Core Team 2012).