Fiscal Multipliers in Denmark1
1. Fiscal policy plays a particularly important role in Denmark given the constraints on monetary policy from the pegged exchange rate and open capital account. Also, the impact of fiscal measures on output tends to be amplified in a fixed exchange rate regime as monetary policy does not respond to offset the output impact of fiscal measures to balance it against other objectives as it would in a regime following a Taylor rule or an inflation target.
2. The IMF’s Global Integrated Monetary and Fiscal Model (GIMF) is used in this chapter to estimate the growth impact of different fiscal measures. The model allows simulations to incorporate the effect of the monetary policy regime in evaluating the impact of fiscal policy. It also allows the assessment of alternative means of maintaining long-run fiscal sustainability to offset the initial fiscal measures (e.g., a reduction in transfers or an increase in labor taxes).
3. Our simulations suggest that the Danish authorities’ planned fiscal consolidation measures for 2015, based on the August 2014 budget outlook, would have a roughly neutral impact on growth.2 In light of the improving output gap, the Danish government planned to consolidate its fiscal position in 2015 through a range of discretionary measures including higher public investment, lower corporate taxes, and pension tax changes. At the same time, the composition of the consolidation package is import so that it will not have too negative impact on growth which could potentially derail the slow recovery. Our simulation further suggests the combined impact of the Danish government’s plans would probably have a positive impact on growth in the short term, despite the overall budget deficit reduction.
4. The chapter is organized as follows. Section B presents an overview of the literature review of fiscal multipliers for Denmark. Section C describes the GIMF model as well as its calibration while Section D presents the results. Section E simulates the growth impact of a given fiscal stimulus with different financing and Section F illustrates a simulation of the current budget plan using the GIMF model. Section G concludes.
B. Literature Review on Multipliers for Denmark
5. Fiscal multipliers for Denmark have been estimated using either econometric estimates or a structural model. The former involves estimating a vector autoregressive (VAR) model. However, the key challenge is to identify structural fiscal shocks from estimated reduced form regressions. In addition, the fact that fiscal multipliers vary across the business cycle makes it difficult to pin point the size of the multiplier at a particular point in time. Alternatively, fiscal multipliers can be estimated using a structural model, in particular, a DSGE model. This type of model has the advantage that agents are forward looking, where, the agents’ decisions today will depend on the expectation of the future. This is particularly relevant, as in reality households’ decisions indeed depend on their outlook of the economy.
6. VAR and other econometric estimates of fiscal multipliers for advanced economies are much higher for spending than for revenue measures. The average first-year multiplier for these countries is around 1 for government spending measures and much smaller for revenue side measures. Model estimated short-run fiscal multipliers are in general smaller than 1. However, they vary significantly depending on the policy instrument used and the openness of the economy, with more openness associated with smaller multipliers (Table 1).
|Source||Methodology||Data sample||Fiscal shock||Country||Fiscal multipliers|
|1 quarter||1 year||2 years||3 years|
|Blanchard and Perotti (2002)||Structural VAR||Quarterly data||Government spending||United States||0.8||0.5||0.5||1.1|
|1960:Q1 - 1997Q4||Taxes||0.7||0.7||0.7||0.4|
|1 quarter||4 quarters||8 quarters|
|Batini, Callegari and Melina||Nonlinear||Quarterly Data||Government spending||Euro area||Expansion||0.4||0.4||0.1|
|(2012)||threshold VAR||1985Q1 - 2009Q4||Recession||2.1||2.6||2.5|
|1 year||3 year|
|Perotti (2005)||VAR||Quarterly data||Government spending||United States||1.3/0.4||1.7/0.1|
|1939Q1 - 2008Q4||1.4/-0.7||23.9/-1.6|
|Ramey (2011)||VAR||Quarterly data||Government spending||United States||0.6 - 0.8|
|1939Q1 - 2008Q4|
|1 quarter||1 year|
|Ravn and Spange (2012)||SVAR||Quarterly data||Government spending||Denmark||1.3||0.5|
|1971Q1 - 2011Q2|
|Barrell, Holland, and Hurst||NiGEM model with||Consumption||Denmark||0.5|
|Christiano, Eichenbaum, and||DSGE model||Government spending||United States||No zero bound is||1.1|
|Zero interest bound is||3.7|
|1 year||2 year|
|Danmarks Nationalbank||DSGE model||Government spending||Denmark||0.8||0|
|1 year||2 year|
|OECD (2009)||OECD global model||Government expenditure||10 OECD||Accommodative||0.9||1.3|
|(consumption and investment)||countries plus||monetary policy|
|Income tax cutrs||euro area||0.6||1|
7. One empirical study focusing on Denmark suggests that the fiscal multiplier of government spending is relatively large but rather short lived. Ravn and Spange (2012) use an SVAR to estimate the fiscal multiplier of government spending for Denmark to be around 1.3. The estimate is towards the upper range of the existing ones for the US which is between 0.8 and 1.5. However, the effect on GDP is only significant for 1 year. This relatively large initial impact reflects the fixed exchange rate framework. In addition, the authors estimate the fiscal multiplier associated with tax measures to be 0.8, where the effect on GDP appeared to be significant for 2 years.
8. A DSGE model for Denmark estimates a multiplier for government spending at around 0.6 on impact. Pedersen (2012) uses the Danish National Bank’s large scale DSGE model to study the effect of a 1 percent of GDP temporary increase in government purchases of goods and services. The increase in spending is initially financed by issuing debt and is subsequently stabilized by means of (lump sum) taxes. The higher government purchases increases real output of around 0.6 percent in the first quarter, which diminishes to zero after 6 quarters. The short-lived nature of the effect could be partly driven by the forward looking agents. In this model, the increase in government debt (due to government expenditure) is assumed to be stabilized by lump sum taxes in the future. As a result, the forward looking agents reduce today’s (private) consumption in anticipation of lower future wealth. In addition, higher government expenditures put an upwards pressure on domestic produced good prices, which leads to a deterioration of competitiveness of Danish firms and in turn dampens the effects of fiscal stimulus.
C. Description and Calibration of the GIMF Model
9. The IMF’s GIMF is a multi-country dynamic stochastic general equilibrium (DSGE) model. It features forward-looking households and firms optimizing their objective functions subject to given constraints. The model also includes frictions such as sticky prices and wages, real adjustment costs, liquidity constrained households, along with finite planning horizons of households, leading to an important role for monetary and fiscal policy in affecting macroeconomic conditions.
10. Both spending-based and revenue-based fiscal measures are non-Ricardian in the model. This allows contractionary fiscal policy to dampen the level of economic activity in the short run while lower government expenditure encourages higher private investment in the longer term. In addition, GIMF enables us to model Denmark as a small open economy with pegged exchange rate to the euro, while allowing the real effective exchange rate to fluctuation depending on the behavior of inflation.
The non-Ricardian features in the model, in order of importance, are:
i. Overlapping generations (OLG) households who value government debt as part of their wealth. These households plan and smooth their consumption over time and their marginal propensity to consume from a stock of wealth depends on taxes or transfers.
ii. Liquidity constrained households who do not have access to capital markets. These households consume their post-tax labor income as well as any transfers they receive from the government in full every period.
iii. Multiple distortionary taxes that affect decisions by households and firms such as: labor income, corporate income and consumption (VAT) taxes.
11. Fiscal policy is modeled using multiple tax and expenditure measures, where the government respects its long-term inter-temporal budget constraint. Government spending can take the form of consumption, investment expenditure or lump-sum transfers, to either all (general) households or targeted towards liquidity-constrained households. Government investment spending augments public infrastructure, which depreciates at a constant rate over time. Taxation includes labor and corporate income taxes as well as consumption (VAT) taxes.
12. We consider a 3-region version of the GIMF model, based on Denmark, the euro area, and the rest of the world. To calibrate the model, we use variety of data sources. For instance, we use national accounts ratios, tax revenues for the different components, general lump-sum transfers, and trade decomposition data for 2012. The debt-to-GDP ratio is chosen to be 40 percent (equal to Denmark’s medium term value). The share of liquidity-constrained households is set to be 25 percent, similar to other advanced economies, however lower than the rest of the world.3 We calibrate the degree of nominal rigidity in Denmark as similar to the euro area, which is 50 percent higher than the rest of the world owing to the persistence of inflation.4
13. We estimate the impact of a permanent 1 percent of GDP improvement in the budget balance through one of seven fiscal instruments. On impact, the fiscal instrument will change by 1 percent of GDP. The government debt level will decrease as a result of the fiscal consolidation as well as the interest rate payment, in turn, this provides additional fiscal space to reverse as much as possible the change in the fiscal instrument being used in the consolidation process. Moreover, we assume that government announcements of any fiscal consolidation are credible, in the sense that households and firms believe the fiscal adjustments are permanent.
D. Fiscal Multipliers in Denmark
14. The estimated multipliers differ across the different measures and over time. The results are presented in Table 2 below with the effect on output over the first five years. It should be kept in mind that even though the consolidation is permanent, the change in the fiscal instrument is temporary because the lower debt level implies lower debt service and interest payments, as explained in the previous section.
|Year 1||Year 2||Year 3||Year 4||Year 5|
|Targeted Lump sum Transfers||0.24||0.19||0.05||−0.14||−0.24|
|General Lump sum Transfers||0.19||0.10||−0.05||−0.24||−0.33|
|Corporate Income taxes||1.00||1.14||1.00||0.90||1.00|
|Labor Income Taxes||0.29||0.48||0.62||0.71||0.81|
|Consumption Taxes (VAT)||0.29||0.29||0.19||0.10||0.00|
15. Fiscal consolidation via government consumption and investment are both associated with higher multipliers although they are larger for investment. Government spending has a direct impact on aggregate demand, while other instruments have to enter through indirect channels on trade, consumption and investment. Government investment has larger and more persistent multipliers due to the loss of productivity from a lower stock of public infrastructure, in addition to the direct effect on real GDP. For both measures, there are two effects offset the impact on GDP: crowding out and leakage. However, given the permanent natural of the shock, the crowding out effect is small. At the same time, because Denmark is a “very” open economy, lower government spending translates immediately to a reduction in import which in turn offset the direct impact of lowering government spending on GDP—leakage effect. In general, our results are in line with the findings by the Danmarks Nationalbank’s DSGE model. However, our estimates suggest that the impacts are more persistent, which is driven by the structure of the model and the underlying rigidities embedded in it.
16. Government transfers to liquidity constrained households or to the general public have lower multipliers in the model. Unlike cuts in government consumption and investment, fiscal consolidation with general transfers does not have a direct impact on aggregate demand. A cut in government transfers targeted towards the liquidity constrained households has a more pronounced and immediate impact on their income, and subsequently consumption, as they consume their present income. At the same time, when government decreases transfers to OLG households, these agents respond by borrowing in the short-run in order to finance their consumption, which they prefer to smooth. Therefore, for general transfers, the impact on output is not large because of the small share of liquidity constrained agents in the economy.
17. On the revenue side, the degree of economic distortion determines the multipliers for the different taxes. Corporate income taxes lead to lower investment and therefore long-run decreases in the level of capital stock and lower capacity. Labor income taxes decrease households’ incentives to supply labor which also reduces the productive capacity of the economy. Moreover, they also reduce the consumption level of the liquidity-constrained households, while the effects on non-credit constrained household are smaller owing to the temporary nature of the tax cut. Consumption taxes have the lowest impact on GDP, as they affect consumption demand and not the factor of production. It is also interesting to note that the effect of higher capital and labor taxes on output accumulates over time as the factor supply is adjusted gradually in response to the tax change. As a comparison, the impact of lower government consumption and transfers is higher in the first year.
18. The estimated effects on GDP are smaller and less persistent if the fiscal consolidation is temporary. A one-off reduction in government consumption of 1 percent of GDP leads to 0.65 percent reduction in GDP on impact but becomes negligible in the second year.
19. Fiscal consolidation has a larger effect on GDP if it is immediately and fully credible. However, the effect of such a policy fades relatively quickly. At the same time, multipliers have smaller but longer impact when economic agents do not perceive fiscal policy as immediate and fully credible.
E. Fiscal Stimulus: Financing Matters
20. This section simulates the growth impact of a given fiscal stimulus with different financing options. We consider a one percent increase in government consumption or investment financed by either a reduction in lump sum transfers or an increase in the labor tax 3 or 5 years after the stimulus package is being implemented. It is important to note that these re-financing options are pre-announced at the time of implementation of the stimulus. In other words, households fully anticipate the stimulus package and that it will be fully financed in the near term.
21. Our simulations suggest that the growth impact is dampened by the anticipated refinancing but varies considerably depending on the measure and timing. As households anticipate the future consolidation, the positive growth impact from the initial stimulus is dampened as household save more. Financing the stimulus with higher labor tax would have the lowest dampening effect on growth in the short run, which will grow over time. This is consistent with our findings in the previous section, where the fiscal multiplier associated with labor income tax is small initially and becomes larger over time. At the same time, financing through lump sum transfers will have a larger and immediate effect on growth which diminishes over time.
|finance in 3 years with …|
|finance in 5 years with …|
|finance in 3 years with …|
|finance in 5 years with …|
F. The August 2014 Budget Outlook in a DSGE Model
22. This section presents a simulation of the discretionary measures proposed by the government for the 2015 budget.5 The simulations are consistent with the August 2014 Budget Outlook and do not incorporate the additional policy measures announced in October 2014. The proposed budget for 2015 includes a mixture of measures, comprising reductions in public investment and lump-sum transfers and increases in public consumption and personal income taxes, implying an overall decrease in the government deficit by 0.4 percent of GDP. These measures can be best mapped into fiscal instruments available in GIMF as follows:
An increase in public consumption by 0.09 percent of GDP;
A reduction in public investment by 0.13 percent of GDP;
A reduction in lump-sum transfers by 0.19 percent of GDP;
An increase in personal income taxes by 0.22 percent of GDP;
A reduction in corporate taxes by 0.12 percent of GDP; and
An increase in VAT (indirect tax) by 0.1 percent of GDP.
23. The GIMF simulations suggest that these combined measures would raise output by 0.05 percent relative to the baseline. The short run growth impact is dominated by the reduction in corporate tax which affects investment positively and has the largest fiscal multiplier attached. However, in the long run, the growth impact is negative as the effects from higher income tax start to kick in and offset the positive short run growth impacts. This suggests that the government has room to consider further consolidation in 2015.
24. The model suggests a ranking of fiscal instruments in terms of their growth impact, where government investment has the largest multiplier. This is the case both on impact and in the medium term. Government investment is followed by government consumption and transfers. On the revenue side, corporate income taxes have the largest multiplier, followed by labor income taxes and consumption taxes.
25. In addition, financing of the fiscal stimulus measures matters, where increasing the labor tax seems to have the least dampening effects in the short run. However, the long run impact can be significant as it takes time for the labor income tax to affect the economy. At the same time, lowering the lump sum transfers has larger dampening effects in the short run, but the effects diminish relatively quickly.
26. A simulation of the proposed discretionary measures for the 2015 budget reveals that the considered measures would have an initial positive effect on output. The combination of fiscal instruments simulated with the GIMF model would increase GDP marginally by 0.05 percent in the first year, suggesting that the government has room to consider further consolidation in 2015. In the medium to long term, the negative effect of higher income taxes would dominate, resulting in a negative effect on output.
BarrelRayDawnHolland and IanHurst.2012. Fiscal consolidation: Part 2. Fiscal multipliers and fiscal consolidations. Working paper 933. Economics Department, OECDParis.
BatiniN.G.Callegari and G.Melina.2012. Successful austerity in the United States, Europe and Japan. Working paper 12.190. IMFWashington, DC.
BlanchardOlivier and RobertoPerotti.2002. An empirical characterization of the dynamic effects of changes in government spending and taxes on output. Quarterly Journal of Economics117: 1329–69.
ChristianoLaurence J.MartinEichenbaum and SergioRebelo.2011. When is the government spending multiplier large?Journal of Political Economy119 (1): 78–121
MineshinaAikoMarcosPoplawski-Ribeiro and AnkeWeber.2014. Size of fiscal multipliers. Post-crisis Fiscal Policy edited by C.CottarelliP.Gerson and A.Senhadji Chapter III.12 pages 315–372. Cambridge, MA: MIT PressJuly2014.
OECD. 2009. The effectiveness of and scope of fiscal stimulus. Economic Outlook (March). Interim report. Paris: OECD105–50.
PerottiRoberto.2005. Estimating the effects of fiscal policy in OECD countries. Discussion paper 4842. CEPRLondon.
PedersenJesper(2012)Fiscal policy in macroeconomic modelsDanmarks Nationalbank Monetary Review 3rd Quarter 2012 Part 2.
RameyValerie A.2011. Identifying government spending shocks: It’s all in the timing. Quarterly Journal of Economics126 (1):1–50.
Prepared by Borislava Mircheva and Jiaqian Chen (EUR).
The government policies announced in October 2014 are not incorporated in these simulations. The government is extending a 2013–14 window to convert certain pension assets from taxable to tax-free status upon withdrawal into 2015 as well as allowing the withdrawal of some pension assets paid in during the 1970s. A discounted rate of taxation applies to incentivize the conversion or withdrawal and generate current tax revenues.
The share of liquidity constrained households does not play an important role in permanent shocks (which are used in our simulations). Liquidity-constrained households cannot save, and hence are forced to consume all of their current income. Therefore, in a forward looking model with permanent shock, these agents are not forced to adjust their consumption as they would be in the case of a temporary shock.
Changing the degree of nominal rigidity does not significantly affect the simulation results.
Budget Outlook 2, August 2014 http://uk.fm.dk/publications/2014/budget-outlook-2-august-2014/