Quantitative easing (QE) has been a pivotal monetary policy tool during periods when short-term interest rates are pinned at the effective lower bound, notably after the Global Financial Crisis and during the COVID pandemic. Despite its widespread use, QE has faced significant criticism in recent years for causing large central bank losses and contributing to the post-COVID inflation surge – in contrast to the post-GFC experience in which central banks made profits and inflation remained subdued.  

This critique of QE, and varied experiences across episodes, raises questions about whether and when policymakers should deploy it to fight the next recession.  Under what conditions is QE likely to be warranted, and when may central banks want to hold back?  Does it make sense to engage in QE to address a ‘lowflation’ problem even though the economy is close to full employment?  And taking stock of the recent inflation surge, does the implementation and communication of QE require modification to reduce potential risks of overheating?  

In assessing the merits of QE, it seems natural to begin by considering its macroeconomic effectiveness – how much it boosts output and inflation – under different recessionary circumstances.  There is less agreement about how to assess the fiscal costs.  But we argue that focusing on central bank losses – as often done in the financial press – is far too narrow a standard.  What should matter from a societal perspective is how QE affects the consolidated fiscal position of the government and central bank together (English and Kohn 2022), though recognising that large central bank losses can potentially weaken central bank independence and hence do warrant some consideration. 

Assessing the joint macro and fiscal consequences of QE using purely empirical models is challenging, both in terms of specifying a sufficiently rich framework, but also in terms of identifying the partial impact of QE conducted by central banks. Hence, our analysis, which we present in more detail in Adrian et al. (2026), relies on a structural dynamic general stochastic equilibrium (DSGE henceforth) model that allows for an empirically realistic account of the key transmission mechanisms of QE, including its effects on term premiums, output, inflation, and government debt.  

In the remainder of this column, we provide a brief overview of our model and then consider the macro and fiscal implications of QE under recessions of varying severity.  We also compare QE to standard fiscal stimulus, which seems particularly relevant in an environment of high public debt where many economies have limited fiscal space.  Finally, we consider some policy implications of our findings.

Analytical framework

Our model is based on the New Keynesian paradigm and combines several features that strengthen its realism and policy relevance. First, it includes segmented bond markets following Chen et al. (2013), which allows QE to affect term premiums and economic activity in a meaningful way. Second, it incorporates behavioural discounting, as in Gabaix (2020), to address the forward guidance puzzle, thereby preventing the model from overstating the potency of announcements about the future path of interest rates. Third, it introduces a nonlinear Phillips curve, following Harding et al. (2022, 2023), which is important for capturing the possibility that inflation can accelerate more quickly once there is less economic slack. The model also includes standard frictions such as sticky prices, sticky wages, and consumption habits. On the fiscal side, the framework tracks both short- and long-term public debt, government spending, taxation, and the profit and loss transfer between the central bank and the treasury.

The model is calibrated to US and euro area data, and it implies transmission of short-term policy rates and government spending shocks in the mid-range of empirical estimates. On QE, our calibration implies effects on the term-premium that are towards the lower end of empirical estimates, while the output effects are on the conservative side.  For instance, our calibration implies that QE equal to 10% of baseline GDP reduces the term premium about 50 basis points, which is below the average in survey evidence reported in Gagnon.  And the output effects are smaller (0.7-0.8%) than in the empirical evidence (1.1%) presented in Fabo (2021) drawing on research outside central banks.  

Main findings

Our core argument is that QE can stimulate the economy and improve the consolidated fiscal position, particularly in deep liquidity traps where the policy rate is constrained for an extended period. 

QE in deep liquidity traps

Figure 1 shows the possible benefits of QE in deep liquidity traps. Panel A shows the transmission of 10% of baseline QE in a prolonged liquidity trap in which the shadow rate falls deeply below the effective lower bound (here, the zero lower bound, or ZLB, for simplicity). In such a situation, QE lowers the term-premium by 50 basis points (not shown) and boosts output by about 0.8% relative to the recession baseline (difference between solid blue and dashed orange line in Panel A).  QE also boosts core inflation by 0.2 percentage points relative to baseline (not shown). Over the horizon shown in the figure, a standard discounted quadratic inflation-output loss function – which is often used as a proxy for welfare – declines by about 20%.

A key question in an environment of high public debt is how QE performs relative to fiscal expansion. In this vein, Panel B compares the effects of the same 10% of GDP expansion of QE (blue dashed lines) to conventional fiscal stimulus that is sized to generate the same output path (solid red lines).  QE lowers the consolidated debt-to-GDP ratio significantly – by approximately 6 percentage points over five years – through higher tax revenues, lower debt servicing costs, debt deflation effects, and raising central bank profits (with the latter reflecting that the yield on the long-term bonds the central bank purchases remains persistently above the short-term policy rate).  In sharp contrast, conventional fiscal stimulus raises government debt by about 3 percentage points of GDP, which mainly reflects that higher government spending has a big ‘direct’ effect in deteriorating the primary balance, and less favourable effects in boosting tax revenue than QE.  

These results point to the attractiveness of using QE in a deep recession, including to help allay the eventual fiscal burden, even while acknowledging that QE has some limits and that fiscal stimulus should accordingly play some role.  Moreover, these results turn out to be quite robust to alternative scenarios in which the economy makes an unexpectedly rapid recovery after QE is deployed. 

Figure 1 QE and fiscal in deep liquidity traps

A) Transmission of QE

B) Comparison of QE with fiscal stimulus

QE in shallow liquidity traps

Figure 2 shows the impact of the same QE programme as in Figure 1, but in a situation when the output and inflation contractions are less severe, so that the shadow rate is expected to undershoot the ZLB for a much shorter period than considered above (see the solid blue lines). QE still has favourable effects in supporting output and inflation in this recessionary environment (the difference between the orange dashed and solid blue lines), even if modestly smaller given that the stimulus exerts more upward pressure on long-term interest rates than in a deep trap.  However, the benefits are much less robust, as QE can intensify overheating if the economy recovers faster than was expected at the time QE was implemented. The green dashed line in Panel A illustrates such a ‘faster recovery’ scenario, where a mix of demand and supply shocks hitting the economy in period 7 (as indicated by the vertical, black dashed lines).  QE clearly amplifies economic overheating and inflation overshooting (reflecting the nonlinear Phillips curve effects in our model).  As highlighted in Adrian (2026), these effects tend to be stronger when QE is combined with commitment-based forward guidance about the policy rate that delays policy rate hikes.

Figure 2 QE in shallow liquidity traps

A) QE under modal outlook and with a faster recovery

B) QE under general shock uncertainty

Notes to Panel B: Modal (no-uncertainty) baseline projection (dashed lines), along with mean (solid black) and the 68th, 80th and 95th confidence intervals from the distribution with QE (20% of baseline GDP), based on 500 simulations with supply and demand shocks that enables the model to match volatilities and correlations for core inflation, output growth, policy rate, and hours worked per capita in the US 1960Q2-2019Q4.

Panel B shows stochastic model simulations that capture future shock uncertainty and embed a larger QE programme (20% instead of 10% of baseline GDP) implemented when the initial term-premium is lower (0% instead of 1%, so riskier for the central bank). Under these circumstances, the left chart in Panel B shows that faster-than-expected recoveries can lead to significant central bank losses due to the crystallisation of duration risk (as the central bank holds long maturity bonds purchased when long rates were low and then faces much higher financing costs tied to the short rate).  Despite these sizeable risks to the central bank balance sheet, the right chart in Panel B suggests that it still is unlikely that QE deteriorates the consolidated fiscal position, as unexpected shocks that lead to sizeable central bank losses are typically not big enough to offset the gains from an improved primary balance and lower debt service costs.  Hence, a key argument against using QE in shallow liquidity traps is not large consolidated fiscal losses, but a risk that it will contribute to overheating if the economy recovers faster than initially envisaged. The use of escape clauses in forward guidance – which would allow the policy rate to rise more quickly if the economy rebounds – could help mitigate this risk to some degree.

Conclusions 

Our analysis suggests significant benefits of using quantitative easing as a tool to fight recessions when inflation is expected to remain well below the central bank’s target for a protracted period. Even so, there is reason for caution when deploying QE in shallow recessions where it runs more risk of causing the economy to overheat.  If QE is used in shallow traps, ensuring flexibility to allow timely policy rate adjustments in response to evolving economic conditions – including through explicit escape clauses – can help limit overheating risks and above target inflation.  

Our work highlights the importance of evaluating the fiscal impact of QE through the consolidated government balance sheet rather than focusing narrowly on central bank profitability.  Even so, given that central bank losses and weak capital positions can potentially weigh on central bank independence, more consideration of the appropriate design of profit-sharing arrangements between central banks and fiscal authorities seems warranted (Adrian et al. 2025). 

Authors’ note: The views expressed herein are those of the authors and should not be attributed to the IMF, its Executive Board, or its management.

References

Adrian, T, C Erceg, M Kolasa, J Linde, R McLeod, R Veyrune, and P Zabczyk (2025), “New Perspectives on Quantitative Easing and Central Bank Capital Policies”, in D Broeders, A Houben, and M Bonetti (eds), Central Bank Capital in Turbulent Times: The Risk Management Dimension of Novel Monetary Policy Instruments, Springer.

Adrian, T, C Erceg, M Kolasa, J Lindé, and P Zabczyk (2026), “Macroeconomic and Fiscal Consequences of Quantitative Easing”, forthcoming in NBER Macroeconomics Annual 2026.

Aguilar, P A, M Alloza, J Costain, S Hurtado and J Martínez-Martín (2025), “The effect of the European Central Bank’s asset purchase programmes on Spain’s public finances”, Bank of Spain Documentos Ocasionales No. 2409.

English, W B and D Kohn (2022), “What if the Federal Reserve Books Losses Because of Its Quantitative Easing?”, Brookings Institution “Up Front” blog, 1 June.

Fabo, B, E Kempf, M Janokova, and L Pastor (2021), “Fifty Shades of QE: Comparing Findings of Central Banks and Academics”, Journal of Monetary Economics 120: 1-20.

Gabaix, X (2020), “A Behavioral New Keynesian Model”, American Economic Review 110(8): 2271-2327.

Gagnon, J E (2016), “Quantitative Easing: An Underappreciated Success”, PIIE Policy Brief PB16-4.

Harding, M, J Linde, and M Trabandt (2022), “Resolving the Missing Deflation Puzzle”, Journal of Monetary Economics 126C: 15-34.

Harding, M, J Linde, and M Trabandt (2023), “Understanding Post-COVID Inflation Dynamics”, Journal of Monetary Economics 140: S101-S118. 

Kolasa, N, S Laséen, and J Lindé (2025), “Unconventional Monetary Policies in Small Open Economies”, IMF Working Paper 25/66.