Presentation is loading. Please wait.

Presentation is loading. Please wait.

9th Euroindicators Working Group

Similar presentations


Presentation on theme: "9th Euroindicators Working Group"— Presentation transcript:

1 9th Euroindicators Working Group
Luxembourg, 4th & 5th December 2006 Eurostat - Unit D1 Key Indicators for European Policies

2 Alternative trend-cycle decompositions
Item VI - 5 of the Agenda Alternative trend-cycle decompositions By Martin Weale Doc 190 / 06 National Institute of Economic and Social Research

3 Purposes To produce trend and cycle estimates for the Euro Area and to make them available on a continuing basis. Production started in October 2006 Consider Industrial Production, quarterly GDP and quarterly employment The estimates are produced five days after the relevant official data appear.

4 Techniques Three methods are used
Unobserved components (Harvey-Trimbur) Christiano-Fitzgerald Filter Hodrick-Prescott Filter These are bound to give different results and the divergence between them may be a useful heuristic guide to the problem of identifying the trend and cycle. We have not used multivariate methods for estimating the cycle (but inflation may provide information on the output trend and cycle)

5 Real time Problems Estimates produced at any point in time depend on future data as well as past data. The estimates of the cycle depend on the forecasts and therefore on the method used to forecast the data. Since forecasts are uncertain the estimates of the trend associated with any particular method are also uncertain. Orphanides and van Norden have suggested that revisions to estimates are of similar magnitude to the estimates of the cycle themselves

6 Uncertainty (i) Like any good data estimates of the cycle should have error margins associated with them. The unobserved components model can conveniently be used to produce estimates of error margins associated with forecasts. Stochastic simulation allows us to produce error margins for the estimates of the cycle at any point in time.

7 Latest estimates for EU25 Industrial Production

8 Latest estimates for EU25 Employment

9 Latest estimates for EU25 GDP

10 Descriptive Statistics: the Industrial Production Cycle, 1990-2006

11 Uncertainty (ii) We illustrate this using industrial production because the sample period is longer than for the other variables. Estimate the cycle over the period 1990m1-2000m1 Produce an estimate of the state of the cycle in 2000m1. Re-estimate recursively for each month up to 2006m8 We use only the final data vintage.

12 Uncertainty bands for EU25 Industrial Production (UC)

13 Simulated real-time versus “final” cycles for EU25 IP(UC)

14 Summary Real-Time Performance Statistics
Harvey-Trimbur Estimates


Download ppt "9th Euroindicators Working Group"

Similar presentations


Ads by Google