Republic of Slovenia: Selected Issues Paper

Republic of Slovenia: Selected Issues


Republic of Slovenia: Selected Issues

ECB Quantitative Easing: Implications for Fiscal Policy1

A. Introduction

1. The past few years have witnessed substantial monetary easing by the ECB. With inflation running well below target, the ECB has been pursuing unconventional monetary policy easing actions, as the zero lower bound (ZLB) is constraining the use of its short-term interest rate instruments. The latest round of quantitative easing (QE) implemented since mid-2015 constitutes the most significant easing episode in the EA to date. The ECB plans to keep this policy in place at least through March 2017, but has not ruled out extending it beyond that date, and/or broadening the range of eligible securities, depending on progress on reaching its inflation target.

2. The latest round of QE focuses on the long end of the yield curve. QE is implemented primarily via purchases of long-term paper of EA sovereigns in the secondary market. The aim is to bring down long-term interest rates, and thereby boost credit (which remains subdued since the global financial crisis) and aggregate demand. Other potential transmission channels include the exchange rate, wealth, and portfolio rebalancing channels.

3. The planned size of QE is substantial, both for the EA as a whole and for Slovenia. With total planned purchases of government paper in excess of €1 trillion (Table 1), the size of QE is very large when compared with earlier ECB easing episodes. By end 2015 roughly ⅓ of the planned purchases had been implemented, with the rest still in the pipeline. Regarding Slovenia, the size of QE is also quite significant as measured against key debt market metrics: at €4 billion, planned purchases of Slovenian paper represent some 12 percent of end-2015 public debt and around ⅔ of the medium- and long-term debt amortization in 2016–17. As with the rest of EA, the bulk of planned purchases of Slovenian paper remain to be implemented.

Table 1.

Eurosystem Sovereign Debt Purchases under PSPP, ex Supranational Debt

(as of Nov. 30, 2015)

article image
Sources: Bloomberg L.P., ECB, and IMF staff calculations.Note:

assuming monthly purchase volume of EUR44 bln.;

including estimated net issuance until end-2016 and current SMP holdings (where applicable) and inclusion of eligible sub-national debt, after application of security issue and issuer limits;

excluding 12% of supranational debt purchases;

partially consistent refers to the maximum amount determined as the lesser of the capital key-based allocation and the still available stock of eligible government debt (e.g., for Germany the latter is binding);

excluding Greece and Cyprus, adjusted for supranational purchases;

“Others” includes Estonia, Latvia, Lithuania, Luxembourg, Malta, Slovenia, and the Slovak Republic.

4. This paper explores the appropriate response of fiscal policy to QE in the case of Slovenia. While the monetary aspects of QE, including the relevant channels through which it can be expected to impact output and prices, have been researched extensively in the literature, the question of the appropriate fiscal policy response has received much less attention. This is surprising given that the impact of QE on public debt service costs is in many cases quite significant, and that fiscal policy can play a role if other transmission channels turn out to be less potent than expected- particularly in an environment of extensive private sector deleveraging. This paper explores this issue for the case of Slovenia.

5. By way of preview of the paper’s policy conclusions, empirical and analytical considerations would argue for a nontrivial fiscal response in Slovenia’s case. While the temporary nature of QE would argue against deviating from medium-term fiscal targets and the much-needed consolidation of current expenditure, some front-loading of public investment would appear justified. This temporary loosening of the primary balance (relative to a counterfactual without QE) should, however, less than fully offset the lower interest costs, so that the overall budget deficit and public debt should still decline. Importantly, the success of this strategy is critically contingent on putting in place a credible medium-term consolidation plan, underpinned by structural fiscal measures.

B. Fiscal “Windfall” of QE

6. The launch of QE has had a substantial impact on Slovenia’s bond yields. Slovenia’s bond yields had already been declining steadily since the bank recapitalization at end-2013, with spreads vis-à-vis core EA countries having narrowed by almost 400 basis points by early 2015; as a result, Slovenia’s yields had reached Spanish/Italian levels just prior to the launch of QE. Since QE launch, Slovenia’s yields have declined further, by almost 150 basis points, broadly in line with most other EA periphery countries (Figure 1). The contribution of lower spreads vis-à-vis the core to this more recent decline was relatively limited, suggesting that Slovenia’s post-QE trends have been part of the broader EA-wide picture.

Figure 1.
Figure 1.

10-Year Government Bond Yield and Spread vs. Germany


Citation: IMF Staff Country Reports 2016, 122; 10.5089/9781475556070.002.A003

Source: Bloomberg.

7. Slovenia’s public debt service costs have declined relatively more than those of most other EA countries. Calculation of the impact of QE on debt service costs relative to a “non-QE” baseline takes into account sovereign gross financing needs over the QE period (2015-17), but also prefinancing at lower interest rates for future years, as well as debt buybacks and other debt management operations. Application of a consistent methodology across EA countries, which considers the maturity structure of the full range of government securities, suggests that Slovenia’s QE-related “fiscal windfall” is comparatively large (Figure 2): at 0.5 percent of GDP per year on average over 2016-17, it is somewhat lower compared to Italy and Ireland, but significantly larger compared to most other EA countries. These differences reflect less the magnitude of sovereign yield decline (as Slovenia is not an outlier in this regard), but mainly Slovenia’s comparatively large near-term rollover needs. In turn, this is a reflection of Slovenia’s market access difficulties since the global crisis and up to late 2013, which forced large (and expensive) short-term borrowing.

Figure 2.
Figure 2.

Selected Euro Area Countries: Estimates of Funding Cost Savings From QE, 2016–17

(Annual average, percent of GDP)

Citation: IMF Staff Country Reports 2016, 122; 10.5089/9781475556070.002.A003

Sources: Bloomberg, ECB, and IMF staff calculations.

8. The comparatively large QE-related “fiscal windfall” enjoyed by Slovenia renders the question of appropriate fiscal response policy relevant. The policy question is to what extent these temporarily lower debt service costs provide room for some primary balance loosening, and to what extent they should be saved. The remainder of the paper takes up this issue.

C. Fiscal Reaction Functions – Empirical Analysis

9. The question of how fiscal policy tends to respond to borrowing cost shocks has been the subject of limited recent empirical work. The typical methodology of choice tends to employ vector autoregressions (VARs) on single-country or cross-country panels – de Groot et. al. (2015) being a recent reference, although single-equation estimation has also been pursued – see, for example, Cizkowicz et. al. (2015). A consensus result in the literature seems to be that, in response to lower borrowing costs, countries tend to loosen their primary fiscal balance, but not by the full extent of the decrease in debt service costs; accordingly, the overall balance tends to improve and the public debt ratio ends up lower after the shock.

10. As a starting point, we estimated a standard VAR to capture historical fiscal reaction functions to cost of borrowing shocks. Specifically, a 5-variable VAR was employed, including as endogenous variables the effective interest rate R (defined as interest payments over last period’s debt stock), the primary balance as a ratio to GDP PB, real GDP growth g, inflation INF, and the public debt ratio DEBT as a predetermined variable. The system was estimated with one lag (as indicated by the Akaike criterion), employing a Cholesky identification for the endogenous variables, on a panel consisting of all 12 EA countries over the full 1998-2015 EMU period. In line with the recent literature, this was an unconstrained VAR, whereby R is shocked by 1 percentage point, and is thereafter allowed to evolve endogenously.

11. The empirical results of the unconstrained VAR are on the whole intuitive and in line with recent work. The estimated impulse responses (Figure 3) suggest that, following a negative 100 basis point R shock, fiscal policy responds by easing, with PB dropping almost 0.3 percentage points below baseline four years after the shock. Nonetheless, the lower PB offsets the impact of the R shock on debt service costs only partially, and as a result DEBT falls, remaining some 2 percentage points below baseline by the end of the five-year horizon. These results are reasonably close to the literature – e.g. de Groot et. al. (2015). With regard to the other endogenous variables, the impulse responses confirm a positive impact on INF (after a very brief “price puzzle” period); on the other hand no significant impact on output growth could be documented (impulse response not shown).

Figure 3.
Figure 3.

Impulse Responses-Unconstrained VAR

Citation: IMF Staff Country Reports 2016, 122; 10.5089/9781475556070.002.A003

Source: IMF staff calculations.

12. While the above results are suggestive, the unconstrained VAR specification may not adequately capture key design features of QE. Letting R evolve endogenously following the initial shock yields a path that exhibits limited persistence, with about half the shock gone by year two. By contrast, QE is planned to remain in place for at least two years.

13. To address these issues, we also estimated a class of constrained VARs, whereby R is restricted to remain at its initial post-shock level over a two-year period. Beyond year two, the pace of unwinding QE (and hence the evolution of long-term rates) will presumably hinge on the response of EA inflation. For the present purposes, two alternative VAR specifications are employed to capture different possibilities regarding the pace of QE unwinding: A “gradual unwinding” specification (Constrained VAR I), whereby R is allowed to evolve endogenously after year two; and a “rapid unwinding” specification (Constrained VAR II), whereby R goes quickly back to baseline by year three and remains at that level for the remainder of the forecast horizon – this scenario could correspond to strong inflation response that leads the ECB to quickly reverse its securities purchases. Other than the constraints imposed on the path of R, the constrained VARs retain the unconstrained VAR specification.

14. The estimated fiscal response under the “gradual unwinding” specification is qualitatively similar to the unconstrained VAR, but the size of the effects is somewhat larger. The impulse responses corresponding to Constrained VAR I (Figure 4) suggest that the PB response to the R shock is more persistent, with PB falling by 0.35 percentage points below baseline by year one, by 0.5 percentage points by year two, and by 0.6 percentage points below baseline by year three. Thereafter, fiscal policy starts tightening gradually, but PB remains at some ½ percentage point below baseline through year five. Despite this easier fiscal stance (in primary balance terms), lower average borrowing costs and (to a smaller extent) higher inflation keep DEBT somewhat lower relative to the unconstrained VAR throughout the five-year forecast horizon.

Figure 4.
Figure 4.

Impulse Responses-Constrained VAR

Citation: IMF Staff Country Reports 2016, 122; 10.5089/9781475556070.002.A003

Source: IMF staff calculations.

15. The profile of the estimated fiscal response under the “rapid unwinding” specification shows intuitively compelling differences. The impulse responses corresponding to Constrained VAR II (Figure 5) suggest a distinctly sharper tightening as QE is unwound. Specifically, like in the gradual unwinding scenario, PB falls by 0.3 percentage points below baseline by year one and by 0.5 percentage points by year two; after that point, however, a much sharper tightening sets in, with PB getting back to baseline by year four, and actually ending up above baseline by the end of the forecast horizon. These differences in fiscal response make intuitive sense: Faced with a rapid unwinding of QE, the fiscal authorities front-load spending to take advantage of low financing costs, but then tighten sharply to get back to their medium-term baseline as borrowing costs bounce back rapidly. Finally, the decline in DEBT is more moderate compared to the “gradual unwinding” case, ending up less than 1 percentage point below baseline by the end of the forecast horizon, as higher interest cost and lower inflation dominate the tighter PB.

Figure 5.
Figure 5.

Impulse Responses-Constrained VAR II

Citation: IMF Staff Country Reports 2016, 122; 10.5089/9781475556070.002.A003

Source: IMF staff calculations.

16. Overall, despite these differences, the estimation results under both QE-relevant specifications paint a qualitatively consistent picture. The following results appear reasonably robust as regards the fiscal response to lower borrowing costs: the primary balance tends to loosen upfront while borrowing cost remain low; however, this primary easing does not fully offset the lower interest bill; as a result, the overall balance improves and the debt ratio falls.

D. Fiscal Reaction Functions – Analytical Considerations

17. While the empirical results discussed above appear reasonably robust, the question remains whether they can be viewed as optimal policy response in models with optimizing household and government behavior. Answering this question is important for the purposes of conducting welfare analysis that can form the basis for extracting plausible policy conclusions. It turns out that a fairly general class of micro-founded theoretical models, which emphasize consumption smoothing, can yield results consistent with the empirical patterns documented in the previous section. By way of illustration, a slightly modified version of the model employed by de Groot et al. (2015) can help clarify the intuition. The main building blocks of the model are as follows:

  • Household intertemporal utility maximization, with households deriving utility both from private consumption and public good provision – the latter being treated as exogenous and subject to a stochastic shock;

  • Household intertemporal budget constraint, with household income consisting of labor income taxed at an exogenous tax rate, subject to a stochastic shock;

  • Interest rate faced by households is an increasing function of total household debt;

  • Government intertemporal budget constraint, with changes in the expenditure and tax fiscal instruments subject to (convex) adjustment costs;

  • Interest rate faced by the government is an increasing function of government debt.

18. By imposing that fiscal authorities have the capacity to commit to a certain deficit path ex ante, this class of models can generate results consistent with the empirical findings of the previous section. Commitment capacity allows a (benevolent) government to follow a time-invariant optimal policy by maximizing household utility subject to resource constraints and households’ first-order equilibrium conditions. Without going into the details of the requirements for a closed-form solution, the setup of the model provides intuition with regard to the empirical findings of the previous section. Higher government expenditure in response to temporarily low interest rates is a direct implication of optimal household consumption smoothing that includes the utility provided by higher public spending. Moreover, a temporarily low interest rate today (implying that it will be higher in the future) leaves the household and government intertemporal budget constraints unaffected, implying that (noninterest) spending will have to be lower in the future to revert to their respective steady-state debt paths.

19. The assumption of perfect commitment capacity, however, is in many cases a strong one. The perfect commitment assumption rules out time inconsistency, which would lead the fiscal authority to re-optimize each period; in such an environment, it is clear that a time-invariant optimal policy cannot be supported. Yet there is strong evidence that time inconsistency problems are pervasive in policymaking, with present bias for public spending in the case of fiscal policy having received considerable attention in the literature. Wide concern about these issues is evidenced by the perceived need to adopt a variety of commitment mechanisms, such as fiscal rules in the case of fiscal policy.

20. In the presence of time inconsistency suitable extensions of the model can restore the analytical basis of the empirical results. To address these problems, the simple model described above can be extended to incorporate endogenous commitment mechanisms which can support an optimal tradeoff between the government’s committing not to overspend against its desire for flexibility to respond to privately observed shocks to the social value of spending. Extending the model along these lines, as suggested by Halac and Yared (2014, 2015), which in turn builds on commitment mechanisms derived by Amador et al. (2006) in a more general context, would be a relatively straightforward option. Such an extension would add a history-dependent dimension to the optimal rule to support commitment, but would otherwise leave intact most of the key features of the simpler model.2 While the extended model’s additional complexity would typically not allow closed-form solutions, a wide range of calibrated parameter values validates the main empirical results of the previous section: a temporary increase in primary expenditure in response to a temporary decline in interest rates, which however does not fully offset the savings in debt service costs, with debt eventually returning to its original steady state level.

E. Policy Implications and Additional Constraints

21. The empirical results and theoretical discussion of the previous sections provides a basis that can underpin policy recommendations on the appropriate fiscal response to QE in Slovenia’s case. Given the temporary nature of QE, there is little ground to deviate from a medium-term objective for the primary balance that safeguards public debt sustainability.3 At the same time, there appears to be scope for some flexibility in the near term that would allow for a transitory increase in public good provision. In terms of quantifying this extra room, the VAR results under the QE-relevant specifications suggest that roughly half of the QE-related interest savings could be spent over a two year horizon; in conjunction with the estimated “fiscal windfall” for Slovenia (Figure 2), this would imply room to raise primary spending by some ¼ percentage points of GDP each year during 2016-17.

22. Public investment appears to be the best candidate to make use of this extra primary spending space. Given the need to adhere to the medium-term fiscal targets, any up-front spending increase would need to be reversed in later years – this would seem to rule out spending categories such a the wage bill, transfers, and, possibly, spending on goods and services, as increases in these categories tend to be politically difficult to scale back. By contrast, public investment (either the EU-funded or the domestic component) is probably best suited to make use of the temporary spending room, given that it is typically part of a multi-year spending plan, and as such its reallocation across different periods should be easier. An additional argument for choosing public investment is that it could help offset any underperformance in private investment relative to the baseline in the near term, whose projected rebound is key for sustaining the recovery.

23. In determining the scope for additional, front-loaded primary spending, a number of potential additional constraints also need to be taken into account. These constraints could not be adequately incorporated in the empirical and analytical models considered above:

  • Consistency with the SGP: Any near-term stimulus would need to be consistent with SGP requirements, notably the constraint of not breaching the 3 percent deficit limit. Assuming that Slovenia adheres to its 2016-17 budget targets, the temporary expansion suggested above would leave it comfortably below this threshold. While the precautionary arm of the EDP also prescribes a minimum annual structural fiscal adjustment, adequate safeguards that Slovenia will adhere to its medium-term fiscal targets, including importantly through structural fiscal reforms of current spending that strengthen credibility (see also below), should allow some flexibility in this regard.

  • Implementation capacity: Availability of public investment projects with high rates of return and administrative capacity to adequately manage them could also be a potential constraint. Calibrating this constraint at the average of the three highest levels of public investment (in terms of GDP) reached over the previous 10-year period suggests that the envisaged temporary public investment increase should be attainable.

  • Avoiding fiscal procyclicality: The models discussed above do not adequately capture cyclical considerations, but it would arguably not be desirable to be providing extra stimulus if it threatens to move the economy significantly above potential. IMF staff estimates of a still negative output gap, together with absence of inflationary pressures and a large current account surplus provide assurance in this regard.

24. It should be emphasized that the suggested strategy of a front-loaded primary spending expansion hinges crucially on a credible medium-term fiscal consolidation strategy. Indeed, it is essential that higher primary spending in the near term not be perceived as signaling that Slovenia’s medium-term fiscal targets could be compromised, as such credibility costs could more than offset any benefit deriving from the suggested near-term strategy. Accordingly, it is essential that any near-term stimulus should be pursued in the context of developing a credible consolidation strategy–underpinned by an adequate institutional framework–that provides strong assurances that fiscal sustainability will be maintained. In this sense, the envisaged near-term flexibility can be viewed as a benefit of medium-term credibility.


  • Amador, M., I. Werning, and G.-M. Angeletos, 2006, “Commitment vs. Flexibility,” Econometrica, Vol. 74, No. 2, March.

  • Cizkowicz, P., A. Rzonca, and R. Trzeciakowski, 2015, “Windfall of Low Interest Payments and Fiscal Sustainability in the Euro Area: Analysis through Panel Fiscal Reaction Functions,” Kyklos, Vol. 68, No. 4, November.

    • Search Google Scholar
    • Export Citation
  • de Groot, O., F. Holm-Hadulla, and N. Leiner-Killinger, 2015, “Cost of Borrowing Shocks and Fiscal Adjustment,” Journal of International Money and Finance, Vol. 59.

    • Search Google Scholar
    • Export Citation
  • Government of the Republic of Slovenia, 2015, “Stability Programme – Amendments 2015,” April.

  • Halac, M. and P. Yared, 2014, “Fiscal Rules and Discretion Under Persistent Shocks,” Econometrica, Vol. 82, No. 5, September.

  • Halac, M. and P. Yared, 2015, “Fiscal Rules and Discretion in a World Economy,” Columbia Business School Working Paper No. 15-72, September.

    • Search Google Scholar
    • Export Citation

Prepared by Ioannis Halikias.


An extension along the lines of Halac and Yared (2015), in particular, also provides a welfare assessment of decentralized versus centralized (e.g. SGP-type) fiscal rules which is relevant for EA countries.


On Slovenia’s medium-term fiscal strategy, see Government of the Republic of Slovenia (2015).

Republic of Slovenia: Selected Issues
Author: International Monetary Fund. European Dept.