Macroeconomic impacts of higher defence spending: a model-based assessment

Prepared by Nikola Bokan, Pascal Jacquinot, Magdalena Lalik, Georg Müller, Romanos Priftis and Rodolfo Rigato

Published as part of the ECB Economic Bulletin, Issue 6/2025.

1 Introduction

This article uses a suite of models to analyse the macroeconomic impact of higher government defence spending. In June 2025 members of the North Atlantic Treaty Organization (NATO) committed to increase core defence and other related spending by a volume unprecedented in recent history. In light of this pledge, we revisit the size of fiscal multipliers and their determinants across a range of ECB macroeconomic models. This article complements previous ECB analyses by highlighting the role of model differences in the quantification of economic effects of public spending.^[1]

Simulations are conducted around an increase in government consumption from 2% of GDP in 2024 to 3% of GDP by 2028 that illustrates a stylised expansion of defence-related purchases (Chart 1).^[2] The increase begins in 2025, reaches its peak by early 2028 and remains at this higher level for ten years, after which it gradually returns to its original level.

Using alternative model settings makes it possible to explore the determinants of fiscal multipliers and inflation estimates. This article looks at the effects of public spending increases on macroeconomic variables under a benchmark setting before studying the transmission under alternative specifications. It shows (i) the importance of private sector expectations about future financial market developments and deficit financing; (ii) the distributional aspects of increased government spending; and (iii) the extent of intra-euro area spillovers, which in turn vary depending on the instrument used or the origin of imports. The model-based assessment is complemented by Box 1, which puts the model-based results into perspective by providing an overview of fiscal multipliers in the relevant empirical literature on defence spending.

Chart 1

Counterfactual path of the increase in government defence spending

(percentage of GDP)

2 Benchmarking fiscal multipliers across models

To cover a wide range of transmission channels, the analysis uses both semi-structural and dynamic stochastic general equilibrium (DSGE) models developed at the ECB. The semi-structural models employed include core projection models, namely the ECB-MC model (Angelini et al., forthcoming) and its euro area counterpart, ECB-BASE (Angelini et al., 2019 and Bańkowski, 2023). We also use a version of the latter that incorporates forward-looking expectations: ECB-REBASE (Adjemian et al., 2024). In addition, we employ two DSGE models. The first is an extended version of the ECB’s NAWM-E model (Coenen et al., 2024), which incorporates a global and regionally disaggregated structure.^[3] The second, which is a newly developed model with household heterogeneity – a so-called Heterogeneous-Agent New Keynesian model, or HANK model (see Kase and Rigato, 2025) – allows us to capture distributional effects more explicitly. We introduce the same government purchase shock into all models and harmonise the simulation modalities to the extent possible. Importantly, monetary policy is assumed to be active across all simulations.^[4]

The average output multiplier of government spending across models is 0.93 over a two-year horizon, although there is substantial heterogeneity. Chart 2 shows the GDP multiplier and the impact on GDP growth and HICP (Harmonised Index of Consumer Prices) inflation for each model. The multipliers shown here are within the range of those in the empirical literature (Box 1). Overall, semi-structural models display higher multipliers than DSGE models. This mostly owes to (i) the absence of expectation channels regarding the financing of the resulting deficit through taxation in the benchmark setting, and (ii) the smaller effects of interest rates on consumption decisions. By contrast, in DSGE models, the additional spending is necessarily financed with a mix of debt and taxes, which leads to stronger crowding-out owing to the effects of taxation on private sector expectations of future disposable income. Semi-structural models also tend to display a weaker role for monetary policy stabilisation. On the contrary, DSGE models feature strong inter-temporal reallocation via the real interest rate channel. The nominal-side transmission also differs depending on the properties of the model, with the HANK model displaying stronger effects on HICP inflation, mostly because inflation is more sensitive to economic cost pressures in the parameterisation used. The global and regional extension of NAWM-E shows weaker effects, and semi-structural forecast models fall somewhere in between. Given the numerous model-specific characteristics that affect the size of the fiscal multiplier, the following sections explore how the transmission mechanism of government spending can vary when relevant channels are altered.

Chart 2

Impact of an increase in government consumption across models

a) GDP multiplier

(relative change in GDP per unit change in government spending)

b) GDP growth

(percentage point deviation from baseline)

c) HICP inflation

(percentage point deviation from baseline)

3 The role of expectations: anticipating financial market responses and future disposable income

Private sector expectations can greatly influence the macroeconomic effects of increased government spending. To understand this interaction, we compare the predictions of two models with different assumptions regarding the expectations formations mechanism. The first model (ECB-BASE) assumes backward-looking expectations. Generally speaking, this implies that economic agents form their expectations on the basis of historical regularities and past information only. By contrast, the second model (ECB-REBASE) assumes that agents are forward-looking and incorporate news about future shocks and economic developments into their current decisions. This expectation formation mechanism is called model-consistent expectations.

Stronger forward-looking behaviour leads to an earlier and faster increase in financing rates, with an adverse impact on investment, but may stimulate consumption via households’ expectations of higher future income.^[5] Chart 3 compares the macroeconomic response with backward-looking (blue lines) and forward-looking (yellow lines) expectations with unchanged tax rates. Under backward-looking expectations, there is consumption and investment crowding-in as the economy gradually becomes aware of the demand and income impulse, with financial conditions tightening only gradually. Under the forward-looking setting, expectations of future policy rate hikes are reflected in long-term interest rates sooner, and the resulting worsening of borrowing conditions adversely affects investment dynamics over the shorter horizon. Generally, consumption decisions are also affected by tighter borrowing conditions, albeit to a much lesser extent than investment decisions. Instead, consumption increases on account of higher expected income.^[6] The prospects of higher income in the future are anticipated in the forward-looking model, which leads to stronger consumption dynamics. In both the BASE and the REBASE model specifications, private consumption and investment are estimated to be crowded out over the longer run as the effects of higher production and incomes start to fade and the financial market tightening causes the economy to contract. Anticipation of the spending pattern in the forward-looking model also leads to mild frontloading in price and wage setting dynamics and, in turn, to a somewhat faster increase in HICP inflation.

The output multiplier is considerably lower if the public deficit is expected to be financed through an increase in labour taxes. The red and green lines in Chart 3 show specifications where the budget deficit is financed via an increase in taxes on labour income, which introduces an anticipation of future tax hikes at the onset of the increase in military spending. Two financing assumptions for the deficit are shown: a partial tax-financing scheme that still leads to a mild increase in public debt and an illustrative extreme case of full financing via higher taxes. In the forward-looking model, households internalise the repercussions for their future disposable income streams and strongly curtail their current consumption. Despite lower price pressures being somewhat beneficial for investment owing to rather limited changes to financial indicators, the bulk of the output multiplier is determined by consumption dynamics. As a result, output multipliers are dampened considerably.

Chart 3

The role of expectations through the lens of ECB-(RE)BASE

(x-axis: quarters; y-axis: deviation from the balanced growth path/baseline)

4 Distributional consequences and the role of marginal propensities to consume

This section explores the distributional aspects of the fiscal stimulus. It employs the HANK model, which features heterogenous households with large and dispersed marginal propensities to consume (MPCs). As is typical in this class of model, and in line with empirical evidence, households at the bottom of the income distribution have higher MPCs than those at the top. Like in the other models, an increase in government spending is associated with an increase in real GDP, despite a crowding-out effect on private consumption. The blue lines on Chart 4 show the implications of the fiscal shock under a benchmark specification of the HANK model. Private consumption and investment are significantly crowded out owing to both higher interest rates and higher labour taxes.

The effects of fiscal policy are amplified when the additional government spending is targeted towards sectors that employ low-income households. The red lines on Chart 4 show the results under this alternative specification. Since MPCs are negatively correlated with income, in this case the additional spending generates an additional stimulus to private consumption. In contrast to the benchmark specification, there is initially a slight crowding-in effect on private consumption. Consequently, HICP inflation is higher than in the benchmark, as are short-term interest rates. This generates a larger crowding-out of private investment, but the fiscal multiplier remains higher than in the benchmark specification.

Chart 4

The role of non-Ricardian behaviour and MPC heterogeneity

(x-axis: quarters; y-axis: deviations from steady state)

The effects of additional military spending on consumption are positive at the bottom of the wealth distribution and negative at the top. Chart 5 shows the distributional consequences of the increase in government spending under the benchmark specification of the HANK model. Each line corresponds to a different group of households in terms of wealth holdings. There is a clear pattern: households at the bottom of the wealth distribution increase consumption, whereas consumption at the middle and top is crowded out. Since MPCs are larger at the bottom, the increase in labour demand and wages coming from the additional government spending tends to stimulate consumption for those households. At the middle and top of the wealth distribution, higher interest rates and expectations of higher future taxes play a larger role, leading to the observed crowding-out.

Chart 5

Effects of additional government spending across the wealth distribution

(x-axis: quarters; y-axis: percentage deviation of consumption from steady state)

5 Backloaded versus frontloaded spending

This section explores how fiscal multipliers vary when additional government spending is announced in advance of its implementation. This is particularly relevant in the case of the current defence-related commitments, as these have been publicly announced and discussed well in advance of the actual spending, which materialises over a longer period. In contrast to social transfers that are related to automatic stabilisers, discretionary defence spending measures are backloaded and gradual.

Backloaded spending leads to substantially lower effects on real GDP. Chart 6 shows fiscal multipliers for pre-announced fiscal shocks in the REBASE and HANK models. Each point on the curve corresponds to the fiscal multiplier of a one-time increase in government consumption as a function of the number of quarters by which its announcement precedes its implementation. Notably, the spending increase is identical – only the implementation date changes. When agents are forward-looking, the fiscal multipliers decrease as the time between announcement and implementation increases. This is because the prospect of increases in taxes and interest rates owing to future increases in government expenditure leads to an immediate crowding-out of private spending.

Chart 6

Dynamic multipliers of pre-announced shocks

(x-axis: number of quarters between spending announcement and implementation; y-axis: long-run present-value multiplier)

6 Cross-country spillovers

This section looks at the cross-country spillovers of government spending and how these can vary depending on key features of the spending. We first benchmark the effects through the lens of the ECB-MC model. Given the workhorse nature of this model, important parameters, such as the import content of government spending, are set to historical averages. We then consider alternative scenarios using the global and regional extension of NAWM-E to provide a more qualitative illustration from alternative sensitivity exercises.

Both government consumption and government investment generate positive domestic real effects, but the magnitude of spillovers across euro area countries varies depending on the import content of the instrument (Chart 7). Government consumption is assumed to rely mostly on domestic production, with an import share of only around 10%. By contrast, government investment has a much higher import share of around 30%.^[7] As a result, in the short to medium run government consumption causes a stronger increase in domestic output than government investment, while spillover effects on other countries remain small – and even turn negative after three years – as the rise in policy rates outweighs the positive trade effects over the longer term. Conversely, government investment generates positive spillovers throughout the horizon, as it stimulates private investment and partially offsets the dampening effect of higher interest rates.

Chart 7

Domestic and spillover effects of government spending during the third year after the shock

a) Real GDP

(percentage deviation from baseline levels)

b) HICP inflation

(percentage deviation from baseline levels)

On the nominal side, government consumption and investment raise domestic inflation and generate inflationary spillovers across countries but do so in different ways. Government consumption is substantially more inflationary, as it directly increases demand without boosting productive capacity. By contrast, government investment raises inflation in the short term but, by strengthening trend productivity, also helps to ease price pressures over the medium term.

Spillovers operate mainly through the trade channel and less so through interest rates, with the net effect on the euro area appearing limited in magnitude. Chart 8 disentangles the transmission channels using the global extension of the NAWM-E model and shows that higher German public spending raises domestic demand, which initially boosts output. However, under the benchmark calibration, higher government demand crowds out private demand over time.^[8] This is because stronger public demand puts upward pressure on prices and wages, raising the cost of borrowing and reducing households’ and firms’ real disposable income and profitability.^[9] At the same time, the increased demand for imports improves the trade balance of Germany’s euro area partners, supporting their exports and output. However, as prices rise in Germany, the real exchange rate appreciates, which dampens German export competitiveness and increases imports further. This real appreciation partly offsets the boost to net exports for the rest of the euro area. For the euro area, the higher aggregate demand and price pressures push up inflation. In response, the central bank raises the policy rate negligibly, suggesting that the impact of this tightening on private consumption and investment – via the traditional interest rate channel – is likely to be limited.

Chart 8

Macroeconomic effects of increased defence expenditure in Germany according to the extended version of NAWM-E

(x-axis: quarters; y-axis: deviations from steady state)

A higher share of imported goods and services leads to stronger spillovers to the rest of the euro area, but the overall effects on the euro area remain similar. For example, if Germany ramps up defence purchases involving foreign equipment or invests in infrastructure projects requiring imported machinery, a larger share of the increased demand leaks abroad, directly boosting exports and production in Germany’s trading partners (Chart 9, yellow lines).^[10] As a result, the trade channel becomes stronger, and spillovers to the rest of the euro area grow larger. However, the overall effects on euro area output are similar to the scenario with spending that is predominantly domestically sourced, as higher production in the rest of the euro area is offset by import leakage in Germany and a general appreciation of the euro area real exchange rate.

Chart 9

The role of the import content of government consumption

(x-axis: quarters; y-axis: deviations from steady state)

Box 1
Fiscal multipliers of defence spending: a short review of the empirical literature

Prepared by Cristina Checherita-Westphal and Laust Ladegård Særkjær

Although the evidence is mixed across the empirical literature, it suggests that military spending can have positive short-term demand effects which then tend to decrease over time. This broadly confirms the model-based effects of the general government spending increase considered in this article, although results are more dispersed across samples and various state dependencies (Chart A). The size of the defence spending multiplier differs significantly across studies, with the shorter-run effects in most cases below or close to 1 – broadly in line with the suite of model results presented above – although values close to 1.5 are found in several studies. Very few studies directly analyse the impact of higher defence spending on private consumption.^[11] Ilzetzki (2025), one of the most recent literature reviews, concludes that there is a consensus that GDP does increase following higher defence spending, but the degree of this expansion and the potential crowding-out of the private sector are debated. He also points out that two meta-analyses on the topic disagree on the conclusions. Alptekin and Levine (2012), a meta-study of 169 estimates of the military spending multiplier, suggests that military spending has positive, but small, growth effects. Conversely, the updated sample of F. Yesilyurt and M. E. Yesilyurt (2019) shows no relationship. Finally, in their meta-analysis of fiscal multipliers, Gechert and Rannenberg (2018) find that military spending tends to have lower average multipliers than general government spending,^[12] a finding corroborated by a recent study analysing the output effects of defence spending in the central and eastern European members of NATO’s eastern flank (Olejnik, 2023).

There is also considerable heterogeneity in the output effects of various components of military expenditure. The growth effects of military research and development (R&D) spending are found to be considerably higher than those of other components and to exceed unity, with evidence of crowding-in of private R&D.^[13] For the other spending categories, and particularly over the medium to long run, higher defence expenditure is usually found to crowd out resources available for productive purposes. Expenditure on wages or military personnel (not directly covered in the model simulation results above) is found to have lower or (in the longer run) even negative growth effects, as such spending may impede productivity in the remaining civilian sectors of the economy (see Chart A and additional findings in Becker and Dunne, 2023).

Chart A

Fiscal multipliers of military spending in more recent studies

(increase in GDP from a one unit increase in defence spending, unless otherwise specified)

Defence fiscal multipliers are found to be state-dependent. The empirical literature usually finds evidence of fiscal multipliers being larger (i) in recessions (bad economic times), when the share of non-Ricardian consumers tends to increase and public spending can crowd in (rather than crowd out) private investment; (ii) when there is ample fiscal space or sound public finances (lower debt, deficit or lower interest rate-growth differential); (iii) for less open economies (less potential for spillovers outside the economy through imports); and (iv) for fixed-exchange rate systems (inactive monetary policy) (see Warmedinger et al. (2015) for a review). Similar results are found for defence spending multipliers in Sarasa-Flores (2025) (Chart A) and in Sheremirov and Spirovska (2022) (for fiscal multipliers of general government spending instrumented by military spending). The importance of state dependency (non-linearities) for good versus bad economic times is also emphasised in the meta-analysis of Gechert and Rannenberg (2018). Together with fiscal fundamentals, state dependency is an important aspect to be considered in the (linear) model-based simulations. This can be done by adequately varying the model parameters depending on the prevailing state.

Empirical studies of defence spending rarely investigate the effects on inflation directly. The few available studies for the United States tend to find a positive short-term effect. Ben Zeev and Pappa (2017) find military spending news to be inflationary, peaking after one year before returning to zero in year two. Looking at state-level effects on inflation of military spending, Nakamura and Steinsson (2014) find no cumulative effect after two years. More recently, Antonova et al. (2025) find that military spending news leads to higher manufacturing prices in the United States (while the impact on consumer price index inflation is not investigated). This effect was larger and more persistent in the post-Cold War period, when the US manufacturing sector shrank.

7 Conclusion

Against the backdrop of planned increases in defence spending in the euro area, a model-based assessment of government spending shocks suggests a positive effect on real GDP growth and a modest effect on HICP inflation, with significant uncertainty surrounding estimates. A persistent but gradual increase in military spending of 1% of GDP over three years is associated with a two-year GDP multiplier of 0.93 and a two-year impact on HICP inflation of 0.07 percentage points on average across different model types. There is substantial model heterogeneity, with output multipliers ranging from 0.42 to 1.13. Four years after the shock, the GDP multiplier remains at 0.93 on average, while HICP inflation increases by 0.2 percentage points on average across the models. Model-implied fiscal multipliers are generally aligned with the empirical evidence on the effects of military spending. We further identify key transmission channels that shape the economic effect of additional military spending: a higher share of investment produces larger spillovers, whereas the private sector anticipating higher taxes and interest rates plays a prominent role in reducing multipliers. Several aspects remain outside the scope of this analysis, in particular state dependency with respect to government indebtedness and bad versus good economic times, which is found to be relevant in the empirical literature. Finally, important aspects of military spending cannot be fully captured in this analysis. Examples include the composition of spending by single countries or sectoral supply and industry-network effects.

References

Adjemian, S., Bokan, N., Darracq Pariès, M., Müller, G. and Zimic, S. (2024), “ECB-(RE)BASE: Heterogeneity in expectation formation and macroeconomic dynamics”, Working Paper Series, No 2965, ECB, Frankfurt am Main.

Alptekin, A. and Levine, P. (2012), “Military expenditure and economic growth: A meta-analysis”, European Journal of Political Economy, Vol. 28, No 4, pp. 636-650.

Angelini, E., Bokan, N., Christoffel, K., Ciccarelli, M. and Zimic, S. (2019), “Introducing ECB-BASE: The blueprint of the new ECB semi-structural model for the euro area”, Working Paper Series, No 2315, ECB, Frankfurt am Main, September.

Angelini, E., Bokan, N., Ciccarelli, M., Lalik, M. and Zimic, S. (forthcoming), “The ECB-Multi Country Model: A semi-structural model for forecasting and policy analysis for the largest euro area countries.” Working Paper Series, ECB, Frankfurt am Main.

Antolin-Diaz, J. and Surico, P. (2025), “The Long-Run Effects of Government Spending”, American Economic Review, Vol. 115, No 7, pp. 2376-2413.

Antonova, A., Luetticke, R. and Muller, G.J. (2025), “The Military Multiplier”, CESifo Working Paper, No 11882, CESifo GmbH, Munich, May.

Bańkowski, K. (2023), “Fiscal policy in the semi-structural model ECB-BASE”, Working Paper Series, No 2802, ECB, Frankfurt am Main, March.

Barro, R.J. and Redlick, C.J. (2011), “Macroeconomic Effects From Government Purchases and Taxes”, The Quarterly Journal of Economics, Vol. 126, No 1, pp. 51-102.

Becker, J. and Dunne, J.P. (2023), “Military Spending Composition and Economic Growth”, Defence and Peace Economics, Vol. 34, No 3, pp. 259-271.

Ben Zeev, N. and Pappa, E. (2017), “Chronicle of a War Foretold: The Macroeconomic Effects of Anticipated Defence Spending Shocks”, The Economic Journal, Vol. 127, No 603, pp. 1568-1597.

Checherita-Westphal, C., Huber, C., Rodriguez-Vives, M. and Müller, G. (2025), “Fiscal aspects of European defence spending: implications for euro area macroeconomic projections and associated risks”, Economic Bulletin, Issue 5, ECB.

Coenen, G., Lozej, M. and Priftis, R. (2024), "Macroeconomic effects of carbon transition policies: An assessment based on the ECB’s New Area-Wide Model with a disaggregated energy sector," European Economic Review, Elsevier, Vol. 167.

Gechert, S. and Rannenberg, A. (2018), “Which fiscal multipliers are regime‐dependent? A meta‐regression analysis”, Journal of Economic Surveys, Vol. 32, No 4, pp. 1160-1182.

Gomes, S., Jacquinot, P. and Pisani, M. (2012), “The EAGLE. A model for policy analysis of macroeconomic interdependence in the euro area”, Economic Modelling, Vol. 29, No 5, pp. 1686-1714.

Ilzetzki, E. (2025), “Guns and Growth: The Economic Consequences of Defense Buildups”, Kiel Report, No 2, Kiel Institute for the World Economy, February.

Kase, H. and Rigato, R. (2025), “Beyond averages: heterogeneous effects of monetary policy in a HANK model for the euro area”, Working Paper Series, No 3086, ECB, Frankfurt am Main.

Nakamura, E. and Steinsson, J. (2014), “Fiscal stimulus in a monetary union: Evidence from US regions”, American Economic Review, Vol. 104, No 3, pp. 753-792.

Olejnik, L.W. (2023), “Economic growth and military expenditure in the countries on NATO’s Eastern flank in 1999-2021”, Working Paper Series, No 2023/2, Eesti Pank.

Ramey, V.A. (2011), “Identifying Government Spending Shocks: It’s all in the Timing”, The Quarterly Journal of Economics, Vol. 126, No 1, pp. 1-50.

Ramey, V.A. and Zubairy, S. (2018), “Government Spending Multipliers in Good Times and in Bad: Evidence from US Historical Data”, Journal of Political Economy, Vol. 126, No 2, pp. 850-901.

Sarasa-Flores, D. (2025), “Buy Guns or Buy Roses?: EU Defence Spending Fiscal Multipliers”, Working Paper, No WP 25-06, BBVA Research, June.

Sheremirov, V. and Spirovska, S. (2022), “Fiscal multipliers in advanced and developing countries: Evidence from military spending”, Journal of Public Economics, Vol. 208, 104631.

Warmedinger, T., Checherita-Westphal, C. and Hernández de Cos, P. (2015), “Fiscal multipliers and beyond”, Occasional Paper Series, No 162, ECB, Frankfurt am Main, June.

Yesilyurt, F. and Yesilyurt, M.E. (2019), “Meta-analysis, military expenditures and growth”, Journal of Peace Research, Vol. 56, No 3, pp. 352-363.