Author: admin

Over the past two decades, China has become the primary target of antidumping and countervailing (AD/CVD) investigations globally, accounting for roughly one-third of all cases, as Figure 1 illustrates. Both the frequency of investigations and the share resulting in duties have increased significantly. By 2020, around 15% of Chinese exports to the US were subject to AD/CVD tariffs.

Figure 1 Trends in global AD/CVD investigations and China’s exposure

Notes: Figure 1 illustrates trends in global AD/CVD investigations from 2000 to 2020. Panel (a) reports the annual count of investigations worldwide. Panel (b) displays the share of these investigations that targeted China. Panel (c) presents the share of investigations against China that resulted in affirmative outcomes (i.e., the imposition of duties).
Source: Temporary Trade Barriers Database (Signoret et al. 2020) and authors’ calculations.

Moreover, subsidised firms are disproportionately affected. Figure 2 shows that the distribution of tariffs for subsidised Chinese exporters is significantly right-shifted compared to non-subsidised Chinese exporters, indicating consistently higher duties. This pattern is robust for both large and small firms.

Figure 2 AD/CVD duty distribution for subsidised and non-subsidised Chinese exporters

Notes: This figure compares the distribution of AD/CVD duty rates between subsidised and non-subsidised Chinese exporters. Panel (2a) presents the overall distribution, where the red solid line represents subsidized Chinese exporters and the blue dashed line represents non-subsidized Chinese exporters. Panels (2b) and (2c) show the distributions separately for large Chinese exporters (above-median assets) and small Chinese exporters (below-median assets), respectively. Firm size is measured using the log of total assets.

Motivated by these facts, in a recent paper (Feng et al. 2025) we examine the impact of Chinese subsidies on foreign AD/CVD investigations by combining comprehensive data on Chinese industrial firms with the global universe of AD/CVD cases against China. We find that higher industrial subsidies lead to more adverse AD/CVD measures at all stages of AD/CVD investigation. Specifically, heavily subsidised products are more likely to receive affirmative AD/CVD rulings, which lead to tariffs. Among affirmative investigations, industrial subsidies also lead to higher tariffs. At the firm level, firms receiving larger subsidies are less likely to be granted firm-specific duties, which are lower than product-level tariffs applied to other exporters of the investigated product. Among those that do receive firm-specific treatment, higher subsidies are associated with higher assigned duty rates.

Retaliatory tariffs offset approximately 25% of the subsidy’s positive impact on firm growth. Researchers estimating the effect of industrial policies should account for the associated increase in AD/CVD duties. Ignoring this subsidy cost creates a downward omitted variable bias because the resulting duties undermine firm growth.

AD/CVD duties become a significant subsidy cost for heavily subsidised firms or those seeking firm-specific rates

We find that a one percentage point increase in a firm’s subsidy rate leads to a 0.16 percentage point rise in the expected AD/CVD tariff for an average firm in the economy (see Table 5 in our paper). Despite this moderate average effect, the highly skewed distribution of subsidy rates means firms in the right tail face significant trade cost. A firm increasing its subsidy rate from the 5th percentile (0%) to the 99th percentile (115%) expects to face an 18 percentage point increase in foreign AD/CVD tariffs.

Moreover, among firms that appeal for firm-specific tariffs, as their subsidy information undergoes scrutiny during the investigation, the tariff increase is substantially larger. We find that a one percentage point increase in the subsidy rate leads to a 2 percentage point increase in the assigned AD/CVD tariff (see Table 4 in our paper). At the same time, the probability of receiving the firm-specific rate falls by 0.7 percentage points (see Table 3 in our paper). Given that the average product-level tariff is 145% while the average firm-specific tariff is 81%, this translates into a 47 percentage point increase in the expected tariff faced by firms potentially eligible for firm-specific treatment.

Tariffs significantly erode the benefits of subsidies

The intended benefits of industrial subsidies on firm revenue growth, employment and productivity are partially offset by increased foreign trade protection. Using an instrumental variable strategy, we compare five-year revenue growth across firms with different subsidy levels, both with and without controlling for AD/CVD tariffs. We find that when tariffs are omitted, a 1 percentage point increase in the subsidy rate leads to a 1.2% increase in revenue (see Table 10 in our paper). When tariffs are included, the same subsidy yields a 1.5% gain. The 0.3 percentage point difference implies that retaliation offsets approximately 25% of the subsidy’s growth effect. Similar attenuation is observed in firm-level employment (17%) and productivity (30%).

Conclusion

Researchers have shown the widespread use of industrial policies in countries such as China (Barwick et al. 2024) and their resulting trade spillovers into other economies (Rotunno and Ruta 2024). While many policymakers and researchers have highlighted the benefits of industrial policy, (Juhász et al. 2023), our findings reveal a hidden cost: under WTO rules, subsidies lead to more adverse outcomes at every stage of AD/CVD investigations. This mechanism helps explain the recent parallel rise of industrial subsidies and trade protection. The resulting tariffs offset about 25% of the subsidies’ positive effect on firm revenue growth. Policymakers aiming to promote exports should consider how subsidy design shapes trade responses – targeting sectors that impose less foreign harm or relying on alternative, non-subsidy tools that are less likely to trigger retaliation.

References

Barwick, P J, M Kalouptsidi and N B Zahur (2024), “Industrial policy: Lessons from shipbuilding”, VoxEU.org, 5 November

Rotunno L and M Ruta (2024), “Trade spillovers of industrial policy”, VoxEU.org, 11 November

Feng, Y, H Li, S Wang and M Zhu (2025), “Industrial Policy and Retaliatory Protection under the WTO: Lessons from China”, available at SSRN.

Juhász, R, N Lane and D Rodrik (2023), “The new economics of industrial policy”, VoxEU.org, 4 December.

In the classic maturity transformation model (Diamond and Dybvig 1983), banks finance long-term, illiquid assets with short-term deposits. This mismatch means that when interest rates rise, the market value of banks’ long-dated assets falls more sharply than that of their short-term liabilities.

Whether these valuation effects, which often remain hidden as ‘unrealised losses’ under standard accounting rules, threaten financial stability depends largely on depositor behaviour. Deposit rates tend to adjust sluggishly to market rates. This creates a ‘deposit franchise’ (the present value of banks paying depositors below market rates) that gains value as market rates rise. As a result, most banks’ deposits behave like low-cost, long-dated liabilities, providing a natural hedge against falling asset values.

Drechsler et al.  (2021) show that banks use this deposit franchise to lengthen the maturity of their assets, and that those with the stickiest deposits lengthen the most, something we also confirm for euro area banks (Rice and Guerrini 2025). However, the franchise is valuable only while deposits stay put. Increases in interest rates simultaneously enlarge the franchise and the unrealised losses on the assets it hedges. If confidence falters, a run can destroy the deposit franchise precisely when it is most needed, a mechanism demonstrated by Egan et al.  (2017) and brought vividly to life by the collapse of SVB in 2023.

The failure of SVB, Signature Bank, and First Republic Bank forced policymakers globally to ask: Are other banks sitting on similar latent losses? And how close are these banks to destabilising runs?

Evidence from the euro area

We examine these questions for 139 euro area banks using detailed confidential data (Rice and Guerrini 2025). The ECB’s rate hikes (July 2022 to September 2023) were comparable to those in the US, and euro area banks had also lengthened asset durations during the low-rate decade.

While rising rates boosted accounting profits, they led to a sharp deterioration in the market value of euro area banks’ assets. We estimate that unrealised losses on loans and bonds held at amortised cost averaged about 30% of book equity by September 2023, peaking at 60% for some banks.

These figures are significantly lower than those of US banks, where losses averaged 75-95% of equity, and roughly 10% of banks had larger unrealised losses than SVB (Jiang et al. 2024, Drechsler et al. 2024).

Structural differences and heterogeneity

One key difference is asset duration: US banks hold three times more long-dated debt securities as a share of total assets than euro area banks (Martín Fuentes et al. 2023).
Since bonds are constantly priced in liquid markets, latent losses on bond portfolios are more evident to investors and depositors than those hidden in loan books. Furthermore, 30-year fixed-rate mortgages dominate in the US, whereas in Europe they co-exist with shorter fixes and variable-rate contracts.

This heterogeneity is crucial. Banks in variable-rate countries saw little impact on loan portfolio values but still benefitted from the rising deposit franchise. Indeed, including hedges, one euro area bank in six saw its mark-to-market net worth rise as rates increased.

The impact varied by bank profile (Figure 1). Losses were largest for smaller retail lenders with long-dated mortgages and limited swap derivatives; larger banks used swaps more actively but had a greater share of losses on bond holdings.

Figure 1 Balance sheet revaluations as a share of pre-rate-hike equity, split by bank size

Notes: The figure shows revaluations for banks split by size tercile based on total assets in 2023. Each bar represents an individual bank. The visualisation highlights that smaller banks (a) often faced larger relative losses on household (HH) loans, while larger banks (b, c) utilised interest rate swap (IRS) derivatives more actively but had more of their significant exposure to bond (AC/FV) revaluations.

By September 2023, when the yield curve on AAA-rated government bonds reached its peak, interest-rate swaps (IRS) had absorbed roughly one-fifth of euro area banks’ unrealised losses. Greater cross-country variation in mortgage terms supports a deeper swaps market in Europe, with pension funds also acting as key counterparties. In the US, by contrast, banks have historically hedged far less (Begenau et al. 2015, McPhail et al. 2023).

The deposit franchise offset a further one-third of euro area banks’ unrealised losses at the peak of the cycle. Banks with less concentrated, retail-heavy deposits benefitted most due to the relative inertia of their depositors. Combined, the two mechanisms hedged, on average, 46% of asset devaluations, though effectiveness varied (Figure 2).

Figure 2 Share of asset devaluations hedged by the deposit franchise and interest rate swap derivatives at the bank level

Notes: The x-axis shows the share of asset devaluations hedged through the deposit franchise (blue), interest rate swap derivatives (yellow), and combined (orange). Across all banks, the average combined hedge was 46%. The y-axis shows the number of banks for each hedging bucket.

Regulatory differences also mattered. The 2018 US deregulation exempted banks like SVB from rigorous stress tests and liquidity rules. In the euro area, all significant banks face the same core requirements and entered the cycle with strong liquidity, bolstered by ECB funding and excess reserves, reducing fire-sale risks to meet deposit withdrawals.

Deposit run simulations: A fragile tail

To assess the risk of an SVB-style collapse, we simulated a deposit run. We calculated the outflow of uninsured deposits required to wipe out each bank’s marked-to-market net worth (including the deposit franchise and swaps).

The results reveal a fragile tail (Figure 3). While the average bank was resilient, by September 2023, a 5% outflow of uninsured deposits would have rendered three banks insolvent on a market-value basis – a potential trigger for a cliff-edge run. A 10% outflow would have pushed 26 banks over this threshold.

While less severe than the US, where simulations showed hundreds of banks at risk, euro area banks were not immune. The failure of one vulnerable bank may have triggered a broader systemic panic under less favourable circumstances.

Figure 3 Change in marked-to-market net worth and percent of uninsured depositors running in order to make a bank mark-to-market insolvent

Notes: The figure shows the change in marked-to-market net worth as of September 2023 as a fraction of banks pre-rate-hike net worth on the x-axis. The y-axis shows the share of uninsured deposit outflows that would have caused the bank to be insolvent on a mark-to-market basis.

Lessons for policymakers

The 2022-2023 monetary cycle highlighted key vulnerabilities for policymakers:

1. The deposit franchise is a double-edged sword. A less concentrated, retail deposit base stabilises funding and allows banks to keep deposit rates low. We find that a higher share of uninsured deposits necessitated larger increases in deposit rates to retain flighty depositors, whereas a higher share of household overnight deposits correlated with lower rate rises.
    However, the pricing power from a sticky deposit base also encourages banks to lengthen asset duration, heightening exposure to unrealised losses and cliff-edge runs. Supervisors should scrutinise banks relying heavily on the franchise while holding long-duration assets with limited hedging.
2. There is a need to complete the banking union with a fully fledged European Deposit Insurance Scheme (EDIS). Our simulation shows that a 5% outflow of uninsured deposits in September 2023 may have pushed three euro area banks into mark-to-market insolvency. Because euro area deposit insurance funds are still organised, financed, and ultimately back-stopped nationally, their perceived credibility and speed of payout differ across member states. A completed EDIS would narrow cross border differences and dampen the first mover advantage of an uninsured run.
3. Cross-sectoral relationships in the swap market require careful monitoring. Swaps absorbed roughly one-fifth of unrealised losses, but the counterparties are often a small pool of pension funds. Counterparty concentrations amplify propagation of systemic risk; the UK’s 2022 liability-driven investment (LDI) episode served as a warning. Macroprudential surveillance should map the network of swap positions, not just individual bank positions. Furthermore, structural shifts, such as the Netherlands’ move from define-benefit to defined-contribution pension schemes, could erode the pool of counterparties for banks as these schemes transfer duration risk to individual savers.
4. Digital finance compresses run durations. Mobile banking, instant transfers, and viral information flows accelerate withdrawals, as the US regional bank turmoil demonstrated. Real-time monitoring of retail flows and reviewing stressed-outflow assumptions in the liquidity coverage ratio are prudent next steps, which the European Banking Authority is addressing (EBA 2025).

The euro area banking system passed a severe interest rate risk stress test without systemic bank runs. This resilience is explained by diversity in asset durations, funding mixes, hedging strategies, and market structures, which limited losses and absorbed shocks. This diversity increases overall resilience but requires monitoring of pockets of fragility. Differences in hedging via the deposit franchise and derivatives leave some euro area banks significantly more exposed than others. 

Open questions remain. How will digital channels reshape depositor inertia? Will pension reforms shrink the counterparts to banks’ swap hedges? And can pan European deposit insurance be established before the next cycle turns?

References

Begenau, J, M Piazzesi and M Schneider (2015), “Banks’ risk exposures”, NBER Working Paper 21334.

Diamond, D W and P H Dybvig (1983), “Bank runs, deposit insurance, and liquidity”, Journal of Political Economy 91(3): 401–419.

Drechsler, I, A Savov and P Schnabl (2021), “Banking on deposits: Maturity transformation without interest rate risk”, The Journal of Finance 76(3): 1091–1143.

Drechsler, I, A Savov, P Schnabl and O Wang (2024), “Deposit franchise runs”, NBER Working Paper 31138.

EBA – European Banking Authority (2025), Monitoring of Liquidity Coverage Ratio and Net Stable Funding Ratio in the EU, 4th Report.

Egan, M, A Hortaçsu and G Matvos (2017), “Deposit competition and financial fragility: Evidence from the U.S. banking sector”, American Economic Review 107(1): 169–216.

Hoffmann, P, S Langfield, F Pierobon and G Vuillemey (2019), “Who bears interest rate risk?”, The Review of Financial Studies 32(8): 2921–2954.

Jiang, E X, G Matvos, T Piskorski and A Seru (2024), “Monetary tightening and U.S. bank fragility in 2023: Mark-to-market losses and uninsured depositor runs”, Journal of Financial Economics 159.

Martín Fuentes, N, L Di Vito and J M Leite (2023), “Understanding the profitability gap between euro area and US global systemically important banks”, ECB Occasional Paper Series 327.

McPhail, L, P Schnabl and B Tuckman (2023), “Do banks hedge using interest rate swaps?”, NBER Working Paper 31166.

Rice, J and G M Guerrini (2025), “Riding the rate wave: interest rate and run risks in euro area banks during the 2022-2023 monetary cycle”, ECB Working Paper Series No. 3090.