Evaluation of sovereign loan guarantees

During the COVID‑19 pandemic, Portugal helped companies by having the government guarantee most of the amount of their new loans. From 2020 to 2021 the state backed roughly €9 billion, covering 80%‑90% of each loan and allowing terms of up to six years.

This paper uses a macroeconomic model calibrated to the Portuguese economy and financial system to compare two scenarios: one with loan guarantees program and one without them. The model shows that guarantees make banks safer, increase credit, and produce a small but positive effect on overall output. Guarantees do not create entirely new lending; each euro of guarantee generates about €0.31 of additional credit while substituting €0.69 of credit that would have been issued anyway.

In the model, the fiscal cost to the government is modest: guaranteeing 1% of total credit for a year costs about 0.005% of national output. Dropping the guarantee fee raises that cost by roughly 65%. Because the scheme mainly helped relatively safe firms, the actual fiscal cost was about 83% lower than if riskier firms had received guarantees.

The model decomposes the impact of guarantees into a ‘capital requirement’ effect, which temporarily boosts firm credit and output but raises banks’ default risk, and a ‘loan insurance’ effect, which reduces banks’ default risk and sustains higher credit and output throughout the program. The ‘capital requirement’ effect is akin to a macroprudential capital buffers’ release, with the ‘loan insurance’ effect offsetting it. Results suggest that the positive effects of a conventional loosening macroprudential policy can be enhanced and the negative effects can be compensated with an insurance on banks’ loan losses.

These results should be interpreted cautiously. The analysis does not capture moral‑hazard costs induced by sovereign loan guarantees, nor does it consider the fiscal impact of other pandemic‑related policies or unrelated economic shocks that could affect firm default.

Click here to go to the paper by Ivan De Lorenzo Buratta, and Tiago Pinheiro.