Forecasting with Financial Market Survey Data
In close collaboration with the Center for European
Economic Research (ZEW) in Mannheim, an examination of the forecasting
behavior of experts with respect to financial markets and macroeconomic
time series is being explored. Since 1991, the ZEW has been collecting
monthly data of expected developments of major financial time series
(interest rates, stock indices, exchange rates, inflation, business cycle
appraisals) by means of qualitative surveys. The surveys are only sent to
so-called experts - employees in finance departments of banks, insurances
and industrial companies. The monthly forecasts are partly broadcasted
via the German TV station N-TV and provide an important decision-making
basis for brokers and stock traders. The publication of the business cycle
appraisals has obtained popularity comparable to the "Geschäftsklimaindex"
from the Ifo-Institue in Munich.
The research project exhaustively deals with the analysis
of the forecasting procedures with respect to optimal quantification of the
qualitative survey data. Thereby forecasts based on quantal response models
are compared to forecasts based on time series models. Further questions
cope with evaluation of the optimal forecast period, detection of break
points in the underlying time series, and rationality of the forecasting
behavior of professional forecasters.