Prozentor  >  Forecasting  >  Economic Data

Economic Data Forecasting

Economic analysis and forecasting are a specialty of our organisation.

The focus is on ongoing observation and analysis of short-term economic developments in Germany and other countries. Particular attention is paid to industry developments, price changes, consumer spending, unemployment figures and core inflation.

We generate forecasts for the upcoming six months and coming six quarters, and analyse economic and recession cycles.

What is this kind of forecasting useful for?

This forecasting is utilised to obtain answers to economic questions, commonly including:

• How strong will economic growth the next quarter?
• What factors are currently influencing economic growth?
• In what economic cycle is Germany now in?
• What will happen to consumer spending levels over the next quarters?

What is special about this type of forecasting?

In contrast to the method of surveying experts employed by the big economic research institutes, we conduct forecasting applying mathematical/statistical time series analysis methods. Forecasts are computed using data from Germany's Federal Statistical Office.

We have solved the practical problems frequently encountered in forecasting:

• Consistency: market share forecasts add up to 100%
  (example: total number of unemployed = unemployed women + unemployed men)
• Aggregation: monthly forecasting corresponds in aggregate with the relevant quarterly forecasts
• Multi-step forecasting: We are able to ensure high forecast quality even for 24-step forecasts. The timely, automated processing involved in our forecasting is designed more around a variety of industry data rather than key economic indicators.

Who uses this kind of forecasting?

Economic forecasting is utilised by manufacturers, advertisers and banks interested in obtaining timely projections for their respective industries.

Procedures and methods

Mathematical/statistical time series analysis methods are used in this kind of forecasting, in specific SARIMAX models.

1. Forecasts are generated using multiple SARIMAX models for every individual time series within an economic data set. Thus multiple forecasts are produced for every time series modelling a wide range of aspects and factors.
2. Model blending is applied as a method to combine forecasts into a single, optimised forecast. This forecast takes account of trends, seasonality, external influences and calendar effects.
3. Frequency blending is applied as a method to ensure that monthly, quarterly and annual forecasts produced fulfil time aggregation conditions, so that period values for monthly forecasts correspond in aggregate to quarterly forecasts, which in turn correspond to the annual forecast.
4. he ability to conduct forecast updates makes it possible to generate updated forecasts within incomplete periods, such as a forecast for the current calendar year updated monthly.

Our forecasting models are reviewed annually to ensure forecasting quality.