The Federal Reserve, the ECB, and many other central banks around the world have reported unprecedented financial losses in recent years as a consequence of the rapid rise in interest rates (Claessens et al. 2025, Humann et al. 2023, Federal Reserve 2026, ECB 2026). The sharp drop in profitability has reignited a debate about whether larger balance sheets have exposed central banks to excessive financial risk. Sims (2016) argues that greater financial risk-taking by central banks “invite[s] political second-guessing” that may ultimately threaten the institutional independence underpinning effective monetary policy. Others contend that such risks must be weighed against the macroeconomic benefits of employing balance sheet policies (Cecchetti and Hilscher, 2024a, 2024b).

Much of this debate focuses on ex-post realised income. Surprisingly little systematic evidence exists on how much financial risk central banks take on ex ante, and what drives this risk-taking. In a recent paper (Bartels et al. 2026), we provide new evidence on these questions. Our most striking finding concerns institutions rather than economics: more independent central banks take more financial risk, and it is precisely these central banks that expand their risk positions when fiscal policy tightens.

Measuring risk before it materialises

Most existing studies on central bank profitability measure central bank income after the fact. We instead measure the risk central banks chose to carry before outcomes were known. To do so, we apply the value at risk (VaR) framework to hand-collected balance sheet data for 18 advanced economy central banks over two decades – to our knowledge the first systematic cross-country exercise of this kind. The VaR measures the potential financial loss on an asset or portfolio over a given horizon that should not be exceeded at a specified confidence level, based on historically observed return distributions. To this end, granularity matters: a ten-year Treasury carries very different risk from a T-bill, and a balance sheet dominated by foreign currency like the Swiss National Bank’s behaves nothing like the Fed’s. Our measure captures these differences in the composition of central banks’ balance sheets, not just their size.

Constructing the VaR measure required collecting over 330 annual financial statements between 1995 and 2016 and coding asset holdings at the most granular publicly available levels, by asset class, maturity, and currency, matched to corresponding market return data, and accounting for income and risk-sharing arrangements within the Eurosystem. We also capture the risk in central banks’ direct lending to commercial banks by estimating the worst-case default rate implied by banks’ CDS spreads. Overall, our measure captures the loss that a central bank’s portfolio should not exceed over a one-year horizon 95% of the time.

The sample period, 1995 to 2016, covers the shift from the pre-financial crisis ways of monetary policy to the era of near-zero interest rates and quantitative easing – that is, precisely the period in which the exposures behind today’s losses were built up. The subsequent period of pandemic, energy shocks, and large deficits raises distinct questions, which we leave to future work; but the mechanisms we document operate whenever policy rates approach their effective lower bound.

A dramatic rise in central bank risk

Our estimates reveal a striking picture. Average central bank balance sheet risk increased from less than 1% of GDP in the mid-1990s to approximately 3% by the end of our sample period. For some institutions, such as the Swiss National Bank, the increase was even more dramatic.

Figure 1 shows the evolution of risk for selected central banks. Balance sheet risk was modest and stable prior to 2008, but then rose sharply as central banks deployed asset purchases and other policies that expanded their balance sheets, either in the pursuit of domestic price stability mandates or with a view to stabilising the exchange rate.

Figure 1 Central bank risk-taking over time 

Notes: Bars indicate the asset-level VaR over a one-year horizon (in % of GDP), whereas the black line indicates the total VaR taking into account diversification effects.

Economic and institutional drivers of central bank risk-taking

What explains the rise in risk-taking? Our analysis points to both economic and institutional factors as the most important drivers. First, central banks expand their balance sheets when interest rates approach the effective lower bound, and financial risk increases sharply as a result. This is not simply because central bank balance sheets become larger, as we show by controlling for total assets. Moreover, the relationship with conventional monetary policy tools is non-linear. A fall in the policy rate from 5% to zero more than doubles estimated risk-taking, from about 0.8 to 2% of GDP. Figure 2 visualises the increase in risk as interest rates approach zero. This pattern suggests that the rise in financial risk was not accidental but a systematic consequence of the macroeconomic environment.

Figure 2 Risk-taking as a function of the level of interest rates

Notes: Red dots show estimated VaR and blue dots show fitted VaR (area represents 95% confidence band) based on the main regression specification in Bartels et al. (2026), both conditional on the short-term policy rate.

Second, we find that central banks take on more risk when fiscal policy is contractionary. This suggests that monetary and fiscal policies substitute for each other in delivering macroeconomic stabilisation. Our data do not allow us to pin down the direction of causality, but in either direction the finding speaks against the ‘fiscal dominance’ hypothesis, according to which governments would pressure central banks into more accommodative monetary policies to ease the financing cost of public spending increases.

Third, more independent central banks – measured using the index of Romelli (2022) – take more financial risk, not less, controlling for the policy rate, the size of the balance sheet, and macroeconomic conditions. This result contradicts the fiscal dominance view, which posits that politically captured central banks are more likely to be pushed into riskier positions. Instead, our evidence suggests that independence enables central banks to focus on mandate fulfilment – ensuring price stability, stabilising output – without being constrained by concerns over potential losses or reduced payments to treasuries.

The sharpest evidence for this interpretation comes from the interaction between independence and fiscal policy. The positive relationship between fiscal consolidation and central bank risk-taking is significant only for central banks with high independence. In other words, it is precisely the most independent central banks that step in most aggressively with balance sheet policies when fiscal policy tightens. As Goncharov et al. (2023) document, many central banks are reluctant to report negative profits; our results suggest that this reluctance may be less binding for institutions with stronger independence.

Implications for today’s debate

Our findings carry important lessons for current debates about central bank losses. First, the financial risks that materialised were not the result of reckless behaviour but a systematic policy response to the macroeconomic circumstances following the financial crisis, namely low interest rates and contractionary fiscal policy. Second, central bank independence enabled an appropriate policy response in this environment. Central banks that were freer from political constraints were more willing to deploy their balance sheets as monetary policy tools.

Overall, our results suggest that tolerating higher financial risks may be the price of allowing central banks to pursue their mandates even under challenging conditions. 

References 

Bartels, B, B Eichengreen, J Schumacher and B Weder di Mauro (2026), “Central bank independence and risk-taking at the zero lower bound”, CEPR Discussion Paper 21570 and ECB Working Paper 3079.

Caballero, D, A Lucas, B Schwaab and X Zhang (2020), “Risk endogeneity at the lender/investor-of-last-resort”, Journal of Monetary Economics 116: 283–297.

Cecchetti, S and J Hilscher (2024a), “Fiscal consequences of central bank losses”, VoxEU.org, 26 June. 

Cecchetti, S and J Hilscher (2024b), “Fiscal consequences of central bank losses”, CEPR Discussion Paper 19088.

Claessens, S, T Glaessner and D Klingebiel (2025), “Strengthening central bank financial resilience in a changing world”, VoxEU.org, 7 April. 

ECB – European Central Bank (2026), Annual Accounts 2025. 

Federal Reserve (2026), Federal Reserve Banks Combined Financial Statements. 

Goncharov, I, V Ioannidou and M C Schmalz (2023), “(Why) do central banks care about their profits?”, Journal of Finance 78(5): 2991–3045.

Humann, T, K Mitchener and E Monnet (2023), “Disinflation policies and central bank finances”, VoxEU.org, 12 July. 

Romelli, D (2022), “The political economy of reforms in central bank design: Evidence from a new dataset”, Economic Policy 37(112): 641–688.

Sims, C A (2016), “Luncheon address: Fiscal policy, monetary policy and central bank independence”, Jackson Hole Economic Symposium, Federal Reserve Bank of Kansas City.