Brief Expertise „Alternative Potential Output Calculations“

Tasks

In the discussion relating to the debt brake its procyclical effect is particularly criticized. The latter stems from, among others, the recalculations of potential output within one year, which use the respective current short-term projections of the federal government as data input. If the economic expectation change from one calculation to the next one, new estimates for potential output emerge. Accordingly, also the calculation of the output gap and the business cycle component of the public financial balance are affected.

For the potential output estimation both data released by the federal statistical office and the short-term projections of the federal government are used as data input. For the medium-term horizon trends are extrapolated in accordance with the unified EU-method. The short-term projections are used to account for the (expected) development at the end of the sample. However, by using the short-term projections also its forecast uncertainty and forecast error enter the potential output estimation. Moreover, the econometric filter methods, used for estimating potential output, assign a high weight to the most recent observation, i.e. the short-term projections, in determining potential output particularly for the relevant years of the public fiscal account planning.

Methods

Within this brief expertise we analyses whether differing data input for the potential output estimation with fewer years of forecasted data can improve the revision properties for the process of fiscal account planning. Based on the differing data input, potential output estimates for the years 2013 to 2019 are calculated and the resulting revisions of potential output and the output gap are analyzed.

Based on the short-term projections of the federal government as well as the current method the following calculation regarding potential output, output gap and the business cycle component are conducted:

• Variant I: current method (benchmark)
• Variant II: only official data (T−1) and extrapolation of the trend according to current method
• Variants III to V: Data until T and sequential adding of forecasted years until T+2

Data and other sources

The client is providing for each projection (annual projection, spring projection, autumn projection) of the years 2013 to 2019 the data input used by the federal government as an electronic excel-file. Moreover, the client is providing the current specifications of its potential output estimation procedure according to the EU-method.

Results

Based on the data provided by the Federal Ministry of Finance we get to the point that the average revision of potential output can be reduced by neglecting short-term projections. The lowest average revision is obtained for variant II, which only uses officially released data. Partly using the short-term projections (variant III) increases the average revision. However, the latter still lies below the one of the benchmark method. In total, the differences between the approaches are small, though. The results appear to be driven by the revision of the TFP-trend, which can also be reduced by abstracting from the short-term projections. For the output gap a different picture emerges. In this case, the average revision is the highest for variant III and reduces when short-term projections are taken into account. For the business cycle component of the debt brake similar results are obtained; the average revision is the lowest for variant I increases if parts or the entire short-term projections are neglected.