Government Bonds Ammunitions for the ECB Quantitative Easing

Abstract

The aim of this paper is to investigate a range of financial techniques and policy strategies for a Quantitative Easing (QE) program by the European Central Bank (ECB) that would rely on Euro area government bonds purchases under various modalities and guises. In particular, we provide an empirical study of different possible QE arrangements, with a special focus on a proposal of a securitization structure that could clear the different sovereign bonds from credit risk and heterogenerity. The securitization framework proposal is compared with a program of Euro-area public debt direct purchases under two alternative scenarios of: i) pari-passu creditor status of the ECB with respect to private investors; ii) preferred creditor status of the ECB. We consider three cases of “un-managed” (fixed composition) direct purchases according to which member states bonds are purchased in proportion to: i) the “capital key” representing equity holdings of ECB’s shares; ii) a “liquidity-key” representing the relative amount of sovereign bonds outstanding; iii) an equally-weighted portfolio of sovereign bonds. We evaluate the impact of various SPV characteristics - such as bond tranche attachment choice - as well as various financial market scenarios. For these choices we provide a risk management analytical framework based on Monte Carlo simulations. This allows us to compute the probability distribution of credit losses under various credit spreads and default correlation scenarios as well as to design appropriate stress testing procedures to gauge the robustness of our proposed GBB QE solutions. In this version of the paper our modelling strategy relies on the Gaussian copula assumption in default risk correlation. Using end of October 2014 CDS market quotes for 10 Euro-area sovereign bond market, as well as an estimate of a Gaussian copula measure based on historical CDS quotes time series, we conclude that the expected loss would be about 75 basis points on a 5 year horizon for a senior tranche bond with 30% attachment (that is absorbing losses on the government bond portfolio beyond 30% of its value). For all practical purpose, such expected loss and the corresponding credit spread value would be similar to the risk of the German Bund as measured at the same valuation date. As for the junior tranche bond credit risk, we reckon credit spreads between 200 and 300 basis points - depending on the underlying assumptions - which would be comparable with Portuguese government bond credit risk at the same valuation date. Our stress test analysis - based on historical data recorded in November 2011 (Euro area sovereign debt crisis acme) - shows that the securitization structure would effectively limit the credit risk increase on the senior tranche to 5-6% of expected losses estimated in our benchmark case on a 5 year horizon. As one should expect, this fairly solid protection to the senior tranche would come at the cost of a huge credit deterioration for the junior bond tranche to an expected loss level of 60%. This implication should make the junior tranche bond a financial product suitable only for sophisticated institutional investors.