ABSTRACT
We empirically analyse the relationship between longer term central bank liquidity support and banks’ exposures to governments, using difference-in-differences panel regressions and propensity score matching on a large sample of banks in the euro area. The research question is whether the liquidity operations, which were introduced to prevent disorderly deleveraging, can also be linked to unintended changes in banks’ asset allocations, in particular to carry trades in government bonds. The results show that unconditional and conditional refinancing operations have a different effect on banks’ government exposures. Unconditional longer-term refinancing operations went together with more carry trades in stressed countries, i.e. banks borrowing more while increasing their holdings of government bonds. In contrast, refinancing operations that were conditional on banks’ lending were not associated with such carry trades, highlighting the benefits of conditionality attached to long-term refinancing operations.
Acknowledgments
We thank an anonymous referee, Carlo Altavilla, Christiaan Pattipeilohy, Andrea Tiseno and participants at the INFINITI conference (Valencia, 2017), the IPAG conference (Paris, 2018), the DNB research seminar (Amsterdam, 2018) and the MPC workshop on banking analysis (Paris, 2019) for their comments on a previous version of the paper and Neeltje van Horen for helpful advice. Views expressed are those of the authors and do not necessarily reflect official positions of De Nederlandsche Bank or the Central Bank of Ireland.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 Most of the borrowings from the previously introduced 12 month operation shifted to the 36 month operation.
2 ‘Stressed countries’ are the economies that suffered the most during the euro debt crisis, i.e. the so-called GIIPS countries (Greece, Ireland, Italy, Portugal, Spain) and Slovenia. The other countries included in our sample are: Austria, Belgium, Germany, Finland, France, Luxembourg, Malta and the Netherlands.
3 Dummies for the duration of the other programmes are included as controls, to isolate the effect of the programme included in the interaction term.
4 We also estimated regressions with macroeconomic variables instead of country time dummies, but this alternative did not change the results substantially.
5 Results are not reported but are available upon request from the authors.