An Exploration of Real-Time Revisions of Output Gap Estimates Across European Countries
Summary (2 min read)
1 Introduction
- The output gap is the extent to which the level of aggregate economic activity exceeds (or falls short of) the economy's productive capacity, i.e. the gap between actual and potential output.
- It is very often the case that these methods provide different results.
- These empirical regularities are of course important for policy analysis.
- According to their fi ndings, the bias in structural balances cannot be systematically corrected as it depends on the state of the economy; (ii) the same happens with the real-time revisions of the changes in the output gaps.
2 Data
- The authors data are taken from the European Commission's real-time output gap database.
- In particular, chart 1 shows that output gap revisions are signifi cant in terms of magnitude with an average change of 0.71 pp.
- In the case of Spain, the revisions during expansions are around 4.4 pp. on average, while the magnitude in recession periods is lower but still large, at -2.0 pp.
- According to these results, while the over-estimation of structural balances advocated by Kempkes (2014) holds during expansions, it becomes an under-estimation during recessions.
- To shed some more light on the factors behind the revisions to real time output gaps in levels, the authors fi rst explore the role played by revisions in GDP growth and potential growth, which are the main determinants of output gap estimates.
REAL AND POTENTIAL GROWTH REVISIONS
- (ΔlnY t t+1 -ΔlnY t t-1 ) = (Data revision) + (Forecast error) where superscripts refer to the year in which the GDP growth chart is released.
- In contrast, any revision between the forecast publication at t-1 and the data release at t+1 is assumed to be entirely due to forecast errors.
- The average revision in GDP growth during recessions is -2.99, which is the sum of an average forecast error of -2.88pp.
- This correlation vanishes when the authors consider revisions in potential growth estimates.
- Indeed, a regression of the three factors confi rms that revisions in GDP growth are able to explain around 40% of the change in output gap revisions and its associated coeffi cient is signifi cant, while the coeffi cient on potential growth revisions is not signifi cant and the corresponding R2 falls from 40% to 2%.
SOURCES OF REAL GROWTH REVISIONS
- This decomposition allows us to quantify the contributions of historical real-time revisions in GDP and potential growth together with the initial output gap revisions.
- In addition, the authors can also decompose the revisions in actual GDP growth into data revisions and forecast errors as described above.
- To be more specifi c, the authors use this decomposition for the case of Spain in Chart 9 to confi rm that forecast errors (brown bar) are the main driver of the magnitude and the symmetric behavior of revisions in output gaps (blue line).
- Note also that the residual (yellow bar) includes not only the initial revision in the output gap -2004 in their data -but also the numerical error due to the logarithmic approximation of the decomposition.
5 Concluding Remarks
- The authors have analysed the size and the cyclical behavior of revisions of the output gap in levels and changes.
- This defi nition of real time is relevant within several fi scal rules such as those used in the Excessive Defi cit Procedure of the European Union.
- Moreover, the authors fi nd that revisions in output gap levels are primarily due to macroeconomic forecasting errors, while data updates play a minor role.
- According to this pattern, while the over-estimation of structural balances advocated by Kempkes (2014) holds during expansions, it turns to under-estimation during recessions.
- Turning to real time revisions in potential growth fi gures, the authors fi nd that they are not only smaller than those of actual GDP, but also symmetric across expansion and recession (i.e. real time potential growth estimates are always larger than the fi nal estimates regardless of the cycle).
Did you find this useful? Give us your feedback
Citations
47 citations
43 citations
42 citations
40 citations
40 citations
References
42 citations
41 citations
39 citations
37 citations
37 citations
"An Exploration of Real-Time Revisio..." refers background or result in this paper
...This evidence is in line with that of Kempkes (2014), who claims that, as a result of the downward biases in real-time output gap estimates, structural balances are always over-estimated in real time....
[...]
...This fi nding casts doubt over the result by Kempkes (2014), who argues that resulting structural balances are always over-estimated in real time because output gaps are downward-biased in real time....
[...]
...Therefore, the type of corrections to realtime estimates of the structural balance suggested by Kempkes (2014) should be contingent on the degree of slack in the economy....
[...]
...According to these results, while the over-estimation of structural balances advocated by Kempkes (2014) holds during expansions, it becomes an under-estimation during recessions....
[...]
...According to this pattern, while the over-estimation of structural balances advocated by Kempkes (2014) holds during expansions, it turns to under-estimation during recessions....
[...]
Related Papers (5)
Frequently Asked Questions (9)
Q2. How does the European Commission infer the output gap?
since potential GDP cannot be directly observed from economic data, the European Commission infers the output gap using a production function approach (see Havik et al., 2014).
Q3. What is the main reason why the output gap estimates are often over-estimated?
In particular, it is often justifi ed that the lack of reliability of output gap estimates should lead to less weight being attributed to the concept of structural balances in respect of the fi scal rules that could be substituted for an expenditure rule in which the ceiling on public expenditure growth would be linked to the evolution of (past and future) GDP growth.
Q4. What are the main implications of the output gap revisions?
The authors also show how revisions in output gaps can be decomposed into revisions in real GDP forecast errors, GDP data revisions and potential GDP, with forecast errors chiefl y responsible for the overall real-time revisions in output gaps and their cyclicality.
Q5. What is the definition of asymmetric revisions?
The authors label as asymmetric those revisions that are positive in expansions and negative in recessions; according to this terminology, symmetric revisions would always be either positive or negative.
Q6. what is the impact of a negative bias on real-time output gaps?
In particular, inthe case of the implications for fi scal policy, on which this paper focuses, if a negative bias on real-time output gaps is confi rmed, this would imply that structural fi scal balances estimated in real time would on average be overestimated (and structural fi scal defi cits underestimated), thus providing an optimistic view of the underlying fi scal position of countries.
Q7. What is the main argument of Kempkes?
This fi nding casts doubt over the result by Kempkes (2014), who argues that resulting structural balances are always over-estimated in real time because output gaps are downward-biased in real time.
Q8. What is the importance of the output gap for policy analysis?
Both levels and changes of output gaps can play an important role in the conduct of monetary policy as an indication of infl ationary pressures.
Q9. What is the reason for the revisions of output gaps estimated in real time?
revisions of output gaps estimated in real time might be due to changes in the modelling techniques employed, which could be based on new theoretical or empirical fi ndings regarding the economy under consideration.