An empirical analysis of budget follow-up in the EU

Budget forecasts are more than the divine effort of future results. They give them shape. To ensure accountability, fiscal policy in a democratic society is based on a detailed draft budget adopted by Parliament. Expenditure programs are presented along with revenue estimates, which are strongly linked to the revolving macroeconomic forecast. If lawmakers have a clear vision for the future, they will expect higher revenues and higher spending plans; If they are prudent, they will formulate more cautious estimates with equally prudent spending plans.

Sound macroeconomic and budget estimates are not only relevant in a national context. They play an important role in the smooth functioning of the EU’s economic and financial union. In a system that combines decentralized budget policy with centralized monetary policy, national fiscal development can spread to other countries and needs to be integrated. That is why the EU’s economic governance framework includes the obligation for each member state to submit its short- to medium-term budget plans to Brussels for joint scrutiny. These plans are called stabilization programs for eurozone countries and convergent programs for those who have not (yet) adopted a single currency. The underlying concept is to identify and possibly correct advance developments that conflict with generally agreed financial rules, the Stability and Growth Agreement (SGP). In the past, the Sustainability and Convergence Program (SCP) had to be submitted to the European level every autumn. Since 2011, these have been submitted between mid-April and late April. The change was intended to strengthen multilateral financial scrutiny by giving the Commission and Council as much time as possible to raise potential issues before the budget plan is adopted in Parliament.

The evaluation of SCP naturally begins with a careful analysis of macroeconomic estimates based on financial planning. Although such pre-assessment is an essential element of peer pressure, predictions are inherently uncertain and will always miss the mark. In order to distinguish misfortune from a potential bias, previous assessments need to be supplemented by more analytical testing over several years. An early example of such an experiment is Strauch et al. (2004), Jonung and Larch (2006) and Beetsma et al. (2009), all point to some systematic gap between good intentions and delivery.

In a recent study (Beetsma et al. 2022), we revisited the problem with a significantly larger dataset covering more years (1998-2020) and countries (all 27 EU countries). In addition to evaluating the statistical quality of government budget estimates, we also analyze the potential roles played by drivers of budget deficits and by national independent financial institutions.

The main result

It is well understood that any macroeconomic and budget forecast is likely to be revised over time. However, in the absence of an unexpectedly large push, these corrections should ideally be of medium size and more importantly neutral. In practice, studies have shown that forecasting errors continue to be large and governments tend to be overly optimistic about their budget plans (Merola & Perez 2013, Debrun & Kinda 2017, Flores et al.¹

In the EU, this could be a case in point. Figure 1 shows the balance of the forecast / estimated budget for a given year at different times (wine). The dark blue diamond depicts the budget plans made a year before the year the budget was drafted. The red diamond is the ‘first-published’ assumption, which is made one year after the question. This measure of budget balance is particularly relevant in the European context because it forms the basis for real-time monitoring by financial authorities and financial markets. The vertical difference between red diamonds and dark blue diamonds is the erroneous prediction of the first disclosure of the budget balance. The patterns are clear: (1) the amendments are large and permanent, (2) the position of the planned budget becomes more ambitious in the years to come, (3) the budget plans are systematically lower than the original estimates, and (4) the former post (final) data. Shows that budget estimates are still revised over time, largely due to systemic changes.

Figure 1 Budget balance estimates for Italy, first published and ex post

Source: Sustainability and Integrity Program.
Note:: Nowcasts are statistics reported in the year of reference, first-release statistics are reported in the year following the reference year, one-year (two-year, three-year) advance forecasts are made in the year (two years, three years) before the reference year And pre-post statistics are the statistics reported for the most recent data vintage reference year.

This pattern exists for other member states such as Belgium or Portugal. However, in contrast to previous studies, our results show that, across all EU sample countries, the budget forecast error is close to zero. This is because of the complete differences across different countries: some show a systematic over-optimism while others follow the path of over-optimism. Needless to say, the forecast errors are irrelevant from the European Union’s point of view because they are washed out overall. Quite the opposite: in a monetary union where national monetary policy-making is firmly in the hands of individual member states, systematic or major policy errors can spread to other countries and affect the union’s stability as a whole.

From a policy perspective, the key question is: What drives these forecasting errors? Data exploration shows that the most important explanatory variable of first-release budget error is first-release (real) GDP growth error (see Figure 2). This seems intuitive because budget estimates are based on economic growth estimates. The surprise of a negative or positive growth will thus have a serious budget impact. Econometric analysis suggests that the first-release error in real GDP due to excessive optimism leads to a budget deficit of one percentage point leading to an average decline of 6 percentage points.

Figure 2 The relationship between growth and budget forecasting errors

Source: Sustainability and Integrity Program.
Note:: Three data points are omitted in the graph, namely, Greece in 2009, Ireland in 2010 and Slovenia in 2013. In these instances, the countries recorded a first-release error in the -9.8%, -20.8% and -12.2 budget balances, respectively. %

To better understand this relationship, a more granular analysis focuses on a decomposition between first-release revenue and expenditure errors in first-release budget balance errors. These errors are further divided into base, increment, and denominator effects (see Table 1).² The overly optimistic real GDP growth forecast leads to higher GDP and spending share of GDP than the project. It is partly explained that revenue and expenditure are expressed as a percentage of GDP. However, decomposition shows that this relationship goes beyond mere ‘mechanical’ stimuli. Revenue follows overall GDP dynamics, when expenditures are not directly aligned with an unexpected deficit in output, which reflects the performance of automated stabilizers. A negative one percentage point surprise in nominal GDP growth changes the balance of budgets through -0.7 and +0.4 percentage points through the impact channel of revenue and expenditure growth, respectively.

Table 1 Average of prediction errors and their components

Note:: To arrive at the overall error, subtract the denominator effect from the sum of the other two effects. There may be a difference due to rounding error.

The analysis further shows how the government adjusted the previous year to the forecast error – a kind of education / compensation effect. If spending is higher than the previous year’s projection, governments aim to lower spending for the next period. This is a welcome response from a budget planning / sustainability perspective. On the other hand, an excess of revenue in the previous period leads to an increase in expenditure in the next period – suggesting that governments spend windfall revenue instead of creating a revenue buffer.

Conclusion

The importance of the first-release GDP growth forecast error for accurate budget projections provides some food for concern for the architecture of the EU and the national financial structure. An institutional format that produces a more accurate and less biased macroeconomic forecast will greatly improve budget planning. The SGP’s ‘six-pack’ reform of 2011 requires that national GDP growth estimates be prepared by an independent body or at least approved by them. The decision not to approve introduces a ‘nuclear alternative’, which independent organizations will use only as a last resort, giving governments a somewhat shaky place in their forecasts. It is therefore advisable that national independent financial institutions be entrusted with the task of directly creating macroeconomic forecasts, as this facilitates truly independent estimates. In fact, this setting is less common in the EU. Regardless of this choice, the analysis indicates that the presence of an independent financial institution responsible for approving or producing the GDP forecast strengthens the legitimacy of the budget plan if the institution enjoys a high media presence, which strengthens and emphasizes its role in public discourse. Evaluate them.

Author’s Note: The opinions expressed in this column do not necessarily reflect the official position of the organizations with which the authors are affiliated or for whom they work.

References

Beetsma, R, M Giuliodori and P Wierts (2009), “Planning to Cheat: EU Fiscal Policy in Real Time”, Economic policy 24 (60): 753-804.

Beetsma, R, M Busse, L Germinetti, M Giuliodori and M Larch (2022), “The Road to Hell Paved with good intentions? An empirical analysis of budget follow-up in the EU ”, CEPR Discussion Paper 17154.

DeBrun, X. and T. Kinda (2017), “Strengthening Post-Crisis Financial Credibility – Financial Council on the Rise. A new dataset “, Fiscal Studies 38 (4): 667-700.

Flores, JE, D Furceri, S Kothari and JD Ostry (2021), “The reliability of public loan forecasts”, VoxEU.org, 28 May.

Guttierez, B. and J. D. Han (2021), “The Process of Monetary Policy in the Countries of the European Union”, Journal of International Money and Finance, Upcoming

Jonung, L & M Larch (2006), “Improving Monetary Policy in the EU: A Case for Independent Forecasting”, Economic policy 21: 491-534.

Larch, M, J Malzubris and M Busse (2021), “Optimism is bad for financial results”, VoxEU.org, 04 November.

Merola, R and J Pérez (2013), “Should the role of preparing budgetary projections be delegated to an independent body?”, VoxEU.org, 01 May.

Strauch, R, M Hallerberg and J von Hagen (2004), “Budget Forecasting in Europe – Track Record of Stability and Convergent Programs”, ECB Working Paper No. 30.

Endnote

1 The projection of the year T before the introduction of the European semester is published in the autumn of the T-1 of the year, while after its introduction, the projection of the T of the year is published in the spring of the T-1 year, based on the moment the SCP is released.

2 ‘Base effect’ captures a variable value update related to a given year when the time elapses due to a change in the new available information or method – i.e. basically a mechanical effect. The ‘growth effect’ is the difference between the nominal increase in revenue or expenditure (in euros) and the nominal increase in revenue or expenditure (in euros) projected from the first-release, with some weight factor. The ‘denominator effect’ is the first-release error with the same weight factor in nominal GDP growth.