Global Economic Barometers
The Global Economic Barometers are a system of indicators that enables a timely analysis of global economic development.
They represent a collaboration between the KOF Swiss Economic Institute and Fundação Getúlio Vargas (FGV) based in Rio de Janeiro, Brazil. The system consists of two indicators: a coincident indicator and a leading indicator. The coincident indicator is a figure that reflects the current state of economic activity (referred to as ‘coincident’ in the chart). The leading indicator, on the other hand, gives a cyclical signal that is roughly six months ahead of actual economic development (referred to as ‘leading’ in the chart).
The two indicators comprise the results of economic tendency surveys conducted in more than 50 countries. The aim is to achieve the broadest possible global coverage. One advantage of economic tendency surveys is that their results are usually available fairly quickly and are not substantially revised after their first publication. The time series underlying the KOF-FGV Barometer are obtained primarily from the data provider Thomson Datastream.
The coincident indicator includes more than 1,000 different time series, while the leading indicator comprises over 600 time series. Cross-correlation analysis is used to decide which individual time series are included in the barometers. This involves correlating the individual time series with a reference series. The reference series used is the year-on-year growth rate of global gross domestic product (GDP), where the individual national GDPs are aggregated at purchasing power parity to form global GDP. A time series is only included in a barometer if it shows a sufficiently high correlation and a suitable synchronisation or lead with the reference series.
The chosen individual time series are then aggregated to the two barometers using the method of partial least squares. The last step is to scale the barometer time series in such a way that the long-term mean equals 100 and the long-term standard deviation equals 10.
The two barometers are calculated once every month – usually during the first week of each calendar month. They are therefore available at high frequency and in a timely manner. The algorithm used to construct the two barometers is re-run each time, thereby ensuring the flexibility of the barometer system. However, the publication of a new barometer vintage may involve revising the existing values. Construction-related revisions and revisions in the underlying time series can both cause discrepancies in historical barometer values, although they usually do not amount to major changes of the cyclical signal.