Do market participants' forecasts of financial variables outperform the random-walk benchmark?

In this paper, we evaluate the forecasting precision of survey expectations of the four financial variables in the Prospera survey commissioned by Sveriges Riksbank - one of Sweden's most important economic surveys. Our analysis shows that the market participants in the survey are able to significantly outperform the random walk for only one horizon and variable, namely the three-month horizon for the repo rate. At the longest horizon for the repo rate, and at all horizons for the five-year government bond yield, the random walk significantly outperforms the market participants. For the exchange-rate data studied - SEK/USD and SEK/EUR - no significant differences in forecasting precision in favour of the survey expectations can be established. We conclude that while the Prospera survey might be informative regarding the market participants' expectations, it does not seem to carry much information about the actual future developments of the exchange rates and interest rates covered by the survey.