When the future changes the past: fiscal indicator revisions
The 2020 pandemic economic shock has led to reassessment of fiscal policy measures in 2018 and earlier, because of faulty measurement of unobserved output gaps and structural balances. The current period of suspension of EU fiscal rules should be used to design a better fiscal framework.
The COVID-19 pandemic and the subsequent economic distress caught everyone by surprise. In late 2019, forecasts suggested some economic growth in Europe and elsewhere in 2020, while current projections suggest a 7.4% drop in European Union GDP in 2020. No forecaster should be blamed for not foreseeing the pandemic.
It would be a rather bizarre situation, however, if a shock in 2020 changed our view of fiscal policy in 2018 and before – or in other words, if the future changed the past. But this is what is happening. The 2020 pandemic shock has led to significant changes in certain estimates for 2018. In November 2019 the European Commission estimated potential output (measuring the equilibrium position of the economy), output gaps (the difference between actual output and potential output) and structural balances (the underlying position of the budget balance if the economy was at potential and there were no temporary fiscal measures). A year later, in November 2020, new estimates were published, which differ remarkably from the 2019 estimates even for 2018 and earlier years.
The changes to past estimates are significant because the structural budget balance has a central role in the EU fiscal framework. EU countries have to reach a medium-term objective (MTO) in terms of the structural balance, and if the structural deficit is greater than that objective, countries must implement fiscal consolidation. Badly estimated structural balances complicate the assessment of the attainment of the MTO, or the adjustment path towards it, and might even trigger fiscal consolidation even if the true (but unknown) structural balance would not require it.
A lot has been said about the uncertainty of these estimates. In this post I look at a new aspect: how the 2020 pandemic shock has changed estimates for 2018 in the practice of the three institutions publishing such estimates: the European Commission, the International Monetary Fund and the Organisation for Economic Cooperation and Development.
A concrete example might help understand this phenomenon. In November 2019, the Commission estimated that the French structural deficit was 2.67% of potential output in 2018, practically unchanged from 2015-2017 and smaller than in 2014. A change in the structural deficit is used as an indicator of fiscal policy effort. Thus, the late-2019 estimates suggested that France achieved some fiscal consolidation from 2014 to 2018. But in November 2020, the Commission’s estimate for the 2018 French structural deficit had increased to 3.11%, higher than in 2014-2017, suggesting that France implemented a fiscal stimulus from 2014 to 2018.
Moreover, between end-2019 and end-2020, the data for the actual French budget deficit in 2018 was revised from 2.53% of GDP to 2.29% of GDP, while the data for the 2014 deficit remained unchanged at 3.9% of GDP. So, if anything, the end-2020 estimates could have indicated more fiscal consolidation from 2014-2018, not a stimulus.
To systematically analyse such revisions, I compare the European Commission, IMF and OECD estimates and exclude countries for which there were large underlying data revisions. Statistical offices sometimes refine their data releases based on more detailed data or changes to statistical methodologies, which lead to data revisions. For example, GDP growth in Malta in 2018 was measured at 6.8% in November 2019, but 5.2% in November 2020. When actual GDP data is revised, it is not surprising that potential output estimates are also revised, which might also lead to output gap and structural balance estimate revisions. I therefore include in my analysis only those countries for which GDP growth revisions were relatively small, less than 0.3 percentage points in absolute terms, and for which estimates are published by all three institutions. There are nine such countries: Finland, France, Germany, Italy, the Netherlands, Slovakia, Slovenia, Spain and the United Kingdom. Table 1 shows the average revisions across these nine countries.
In all three sets of estimates, the average revision in actual output growth in 2018 is rather small, since this was a criterion for the selection of the countries. The average revision in the 2018 actual budget balance is also minor, especially in the European Commission and OECD estimates.
But the Commission and OECD estimated much higher output gaps and much lower structural balances for 2018 in their late-2020 estimates than in their late-2019 estimates. In contrast, the IMF estimates for 2018 hardly changed. The Commission also made considerable downward revisions for its estimate of the 2018 potential rate of growth, while the other two institutions made less significant revisions.
The likely reason for the close to zero IMF revisions for 2018 is the use of a different potential output model (Figure 1). IMF estimates attributed part of the 2020 actual output decline to a sudden decline in the level of potential output, which is expected to bounce back in 2021. Most likely, IMF modellers considered that lockdown and social-distancing measures prohibited the full use of available capacities and thus temporarily reduced the economic potential in 2020. In contrast, the European Commission and the OECD modellers might have considered that even these idle capacities constitute part of the economic potential.
The revisions made by the European Commission and OECD are called ‘pro-cyclical’ revisions in economics jargon. An estimate is pro-cyclical if it varies in line with the economic cycle. For example, a potential output estimate for a particular year is pro-cyclical if potential output is seen to be higher in economic booms than in economic downturns.
Because of the coronavirus pandemic, GDP growth forecasts were massively revised downward for 2020. As I have argued, this is quite reasonable since the pandemic was an unforeseen external shock that has impacted European economies dramatically. However, for methodological reasons (the use of various smoothing algorithms and economic forecasts), the potential level of output was revised down not just for 2020, but for earlier years too, and thus the output gap estimates for 2018 and earlier years are now higher than a year ago (same level of actual output minus a lower level of potential output). This in turn led to revised structural balance estimates, leading to strange results, such as the French example noted above: instead of a fiscal consolidation in 2014-2018 as suggested by late-2019 estimates, late-2020 estimates suggested that France implemented a fiscal stimulus in 2014-2018.
Earlier research found pro-cyclicality in estimates of potential output in the EU after the Great Recession. A vicious circle might have been at work: low GDP growth after the Great Recession was seen as structural, so that potential output estimates were revised downward. This led policymakers to believe that further fiscal policy adjustments were needed. The successive rounds of fiscal contractions might have caused further reductions in potential output that validated the initial pessimistic estimates.
European fiscal rules are currently suspended. When they are implemented again, they might provoke premature fiscal consolidation because of the faulty measurement of unobserved output gaps and structural balances, as they did after the Great Recession. To avoid this, the current period of suspension of EU fiscal rules should be used to design a better fiscal framework, to apply when the pandemic-induced suspension ends.
Darvas, Z. (2021) ‘When the future changes the past: fiscal indicator revisions’, Bruegel Blog, 5 January
Republishing and referencing
Bruegel considers itself a public good and takes no institutional standpoint. Anyone is free to republish and/or quote this post without prior consent. Please provide a full reference, clearly stating Bruegel and the relevant author as the source, and include a prominent hyperlink to the original post.