Despite three generations of the Stability and Growth Pact (SGP) rules, the EU rules have failed fully to meet their objectives. Recent debate in Europe has focused on increasing the role of an expenditure rule in the EU framework (e.g. Beetsma et al. 2018; Darvas et al. 2018, and Bénassy-Quéré et al. 2018).
This would restrict spending to grow in line with the growth of sustainable output, adjusting for a debt brake element:
Government spending growth = Real potential output growth + inflation – debt brake term
In this column we look at the performance of the EU’s existing Expenditure Benchmark from the perspective of policymakers aiming to stabilise the cycle in real-time by following a spending rule. We suggest a real danger that faulty potential output estimates lead to procyclical policy.
The central problem we argue is the inherent difficulty of estimating potential output in real time due to its non-observability, unknown structural breaks, and the non-computability of such a complex variable. Policymakers face an important trade-off between putting too much weight on recent developments, leading to them to follow the cycle, and too little weight on current observations, risking missing structural shifts in the economy.
The rationale for a spending rule
Expenditure rules to manage spending are potentially attractive for a number of reasons. First, spending is directly observable, unlike the structural budget balance (Bénassy-Quéré et al. 2018). Because the structural balance is derived by subtracting the estimated cyclical component from the observed balance, any error in the measurement of the cycle or the relationship with revenues (‘revenue buoyancy’) is attributed to the structural component. Second, it is easier for the government to control spending execution than revenue outturns during a given year. Third, over time, the rising costs of providing government services due to lower productivity growth in the public sector and demographic pressures make this a key variable for the government to manage. Fourth, spending rules are closely related to multiannual expenditure ceilings, a set of maximum limits on aggregate and/or departmental spending set out for several years ahead, which are widely advocated as a useful tool for sound budgetary management.
However, expenditure rules still rely on an estimate of potential output growth. Many of the papers advocating the Expenditure Benchmark do not focus on the measurement of potential output growth. Indeed, even the European Commission (2017) proposal to strengthen fiscal responsibility only refers to the “underlying government expenditure path” without specifying how this should be determined.
There are some plausible reasons why potential output growth rate estimation is less problematic than for the output gap. Claeys et al. (2016) note that: “The revisions of the real-time estimates of the medium-term average potential growth rate … were smaller than the revisions of the change in the structural balance estimates”. This is because, for a given year, revisions to potential output growth just reflect that year, while for the output gap the revision depends on the cumulative path of revisions across all years. However, comparisons between the output gap-based and potential output-based approaches do not give a sense of how effective expenditure rules are in absolute terms.
How an expenditure rule works in real-time: EU experience
To shed light on how expenditure rules work in real time, we take data from the European Commission’s CIRCA website on its past forecasts of both potential growth rates and actual output growth rates for 15 member states for 15 years (2004–2018) using the spring forecast vintages released by the European Commission.
The Commission’s estimates of potential output estimates are produced under the EU Commonly Agreed Methodology (CAM). These have a number of problems in the way potential output is estimated and the way projections are constructed. This includes issues around estimation of the Phillip’s curve (Darvas and Simon 2015), the treatment of net migration, the use of actual rather than trend capital stocks, and the reliance on filtering methods. A less widely appreciated issue is that variables are projected further beyond the official forecast horizon as a part-solution to the end-point bias problem, but this is typically done in quite a crude way such as by projecting forward the current growth rate, implying a procyclical pattern.
Two exercises can shed light on the procyclicality of these estimates of potential output. Following the EU Expenditure Benchmark, the allowable rate of expenditure growth is calculated using CAM-based estimates of potential output and the ten-year averaging process that the rules use to mitigate procycality.
First, Figure 1 shows rate of spending growth that would be allowed if based on the current vintage of estimates of potential output. This closely follows the ten-year average for actual real GDP growth rates rather than getting at a more meaningful approximation of ‘sustainable’ growth rates. There is clear pattern that follows the cycle. In many cases, such as in Spain, Portugal, Finland, Greece, and Ireland, the range between the peak and the trough in allowable growth rates range from 2.8 percentage points to almost 7 percentage points (the range is wider still using real-time estimates). This is far from a steady growth rate for spending.
Figure 1 Procyclicality of allowed growth rates under the fiscal rules (current estimates)
Sources: European Commission (Autumn 2017 estimates); authors’ workings.
Note: Data show the implied Reference Rates based on ten-year averages of the estimated potential output growth rates, which are derived using the commonly agreed methodology.
Second, we look at whether revisions to potential output growth rates positively respond to revisions to actual growth rates. If potential output is revised up to reflect stronger cyclical developments, the expenditure rule will be procylclical. We regress potential output revisions on the revisions to real GDP following output (real GDP) revisions observed in a panel regression following Fatás (2018):
Revision to potential output(i,t) = α + β Revision actual output(i,t) + ε
There is a significant positive relationship between revisions to estimated potential output and developments in actual output. We estimate that the average pass-through of actual output to potential output is about one-third for EU countries (meaning that a one percentage point revision to actual output growth leads to about a 0.3 revision to potential output in the same direction). These revisions tend to be correlated, and so build up over time.
Using the 10-year averaging rule mitigates these effects, so that allowable spending growth responds by between 0.15 and 0.5 to revisions in actual output as shown in Figure 2. These results are robust to the exclusion of the financial crisis period, expansions versus recessions, and we control for country and time fixed effects.
Figure 2 Sensitivity of potential output estimates to actual output
Estimated coefficients of percentage point revisions to potential output growth rates
Sources: AMECO; authors’ own calculations.
Note: Coefficients on individual country regressions are shown. All are statistically significant at 95 per cent level of confidence. Standard error bands are shown for 95 per cent confidence interval.
To understand the implications for policy, we take these estimates of the sensitivity potential output and apply them to typically observed forecast revisions (González Cabanillas and Terzi 2012). We also assume a positive pattern of upward revisions during a typical five-year economic upswing. This gives a measure of how much the level of spending under the expenditure benchmark would be revised up over the course of an upswing.
For a country like Ireland, allowed spending drifts up by around 2½% relative to the original level. Finland, Greece, Spain, and Belgium all see departures of 1% or more.This gives a clear sense of how the effects of procyclicality in potential output measurement can compound over time to lead to wide divergences from more sustainable growth rates in underlying government spending.
Keeping government spending on a steady path consistent with the sustainable growth of the economy is a key objective for macroeconomic stabilisation. Expenditure rules are an attractive way of doing this. However, as we have suggested in this column, good estimates of potential output are critical to the success of this approach.
The data-driven assessment of how the rules play out in practice shows that implementing the Expenditure Benchmark based on the EU Commonly Agreed Methodology would lead to spending that is procyclical and prone to significant revisions in the same direction as actual growth. Institutionalising procyclicality in the fiscal rules would only serve to exacerbate the already-procyclical fiscal policy stances that are evident in many countries
Policy solutions include revising the CAM for estimating potential output or, more plausibly, switching to alternative methods that better capture the cycle.
National independent fiscal institutions (IFIs) could play a useful role in developing and validating measures of potential output that more accurately reflect country circumstances, including through work on medium-term forecasts and on ways of estimating underlying potential.
This is likely to require a suite of models approaches as well as the use of judgement, including in trying to identify temporary and permanent shocks. Some methods are better designed to identify shifts in potential than the CAM and can be better adapted to country circumstances, both for estimating potential growth rates and the future path of potential output. These included methods that bring in information from the current account and financial variables, as well as more structural models of the economy. Forecasts for potential at a five-year horizon should be published on a regular basis to help their evaluation and improve performance. The most appropriate model and how much weight to put on the most recent observations is likely to vary over time, suggesting that judgement may be important.
The issue of accurately assessing the path of potential output should be at the centre of fiscal policy frameworks, while also recognising that this is necessarily a challenging task and that policymakers need to try to learn more about the underlying nature of the economy and about structural changes to it. A more robust approach would provide a firm foundation for any fiscal framework, including an expenditure-based rule.
Beetsma, R, N Thygesen, A Cugnasca, E Orseau, P Eliofotou, and S Santacroce (2018), “Reforming the EU fiscal framework: A proposal by the European Fiscal Board”, VoxEU.org, 26 October.
Bénassy-Quéré, A, M Brunnermeier, H Enderlein, E Farhi, M Fratzscher, C Fuest, P-O Gourinchas, P Martin, J Pisani-Ferry, H Rey, I Schnabel, N Veron, B Weder di Mauro and J Zettelmeyer (2018), “Reconciling risk sharing with market discipline: A constructive approach to euro area reform”, CEPR Policy Insight No 91.
Casey, E, and S Barnes, (2018), “Overcoming Procyclicality in the EU Spending Rule” in European Fiscal Board (2019), Independent Fiscal Institutions in the EU Fiscal Framework.
Claeys, G, Z Darvas and A Leandro (2016), “A Proposal to Revive the European Fiscal Framework”, Bruegel Policy Contribution 2016/27.
Darvas, Z, P Martin and X Ragot (2018), “European Fiscal rules Require a Major Overhaul”. Bruegel Policy Contribution 2018/18.
Darvas, Z and A Simon (2015), “Filling the Gap: Open Economy Considerations for More Reliable Potential Output Estimates.” Bruegel Working Paper 2015/11.
European Commission (2017), “Proposal for a Council Directive laying down provisions for strengthening fiscal responsibility and the medium-term budgetary orientation in the Member States”. COM(2017) 824 final.
Fatás, A (2018), “Fiscal Policy, Potential Output and the Shifting Goalposts”, paper prepared for the “The Euro at 20” conference organised by the IMF and Central Bank of Ireland, Dublin, 25-26 June.
González Cabanillas, L and A Terzi (2012), “The Accuracy of the European Commission’s Forecasts Re-examined”. European Economy Economic Papers, No. 476, European Commission.