Sovereign spread divergence owing to inflation and redenomination risk countered by unconventional monetary policy in the Eurozone
Description
Using a panel vector error correction (VECM) model, we assess sovereign spread divergence in the Eurozone. Recent literature defines redenomination risk as a member state’s unilateral exit from the Eurozone, estimated by the quanto Credit default swaps (CDS) spread (qCDS). We show that sovereign spread divergence is driven not only by the qCDS but also by inflation and US dollar liquidity shocks. We also find evidence that the balance sheet structure better describes the European Central Bank’s (ECB) unconventional monetary policy (UMP) shocks than the shadow rate. Such policy shocks had a greater impact in the early 2020s than between the global financial crisis and the COVID-19 pandemic (2008Q1–2019Q4). Thus, this paper makes a new contribution to the debate on the ECB’s instruments to maintain price stability objectives and avoid divergence of sovereign bond yields while emphasizing cross-central bank swap lines.
Files
Steps to reproduce
1. Open Eviews 13, to open ‘Econ_Mod_data.xlsx’, ‘input_level’ sheet, use file\import\import from file menu, select file and sheet, with the basic structure as ‘Dated panel’, where panel identifier series for cross section ID series: ‘country’ and date series: ‘time2’ are. 2. Set time sample to ‘2008q1 2022q3’ for the entire and to ‘2008q1 2019q4’ for the truncated model. 3. Panel VECM model estimation: select quick\EstimateVAR menu, then Var Specification Method: ‘Vector Error Correction’ with endogenous variables: ‘log(VIX_STOXX50E) cbs_5y share_from_debt quanto_cds cpi lsfx ecb_shadow_rate sovereign_premium’, exogenous variables both long-run and short-run: ‘c dummy_esm dummy_recession_ez dummy_eurozone_ms’, set lag intervals for diff. endog.: 1 4 quarters. Then in View\Cointegration test… menu select ‘Case 4a (Johansen-Hendry-Juselius).’ It gives us that 5 cointegration equations are for both for ‘2008q1 2022q3’ and ‘2008q1 2019q4’. Then select Estimate to go back ‘VAR specification’ and in the Cointegration menu set the number of cointegrating equations to 5 with the deterministic trend specifications of 4a. 4. For model stability test, please select View\Lag structure\AR Roots table or AR Roots graph, where VEC specification imposes 3 unit roots, since the number of endogenous variables are 8, there are 5 cointegrations, therefore there supposed to be 3 Modulus with 1.0000 as a value. This is how “Figure 1: AR graph for stability check for the entire period (left) and pre-COVID-19 period (right)” was generated. 5. Then select again View\Cointegration test… menu select ‘Deterministic assumptions: Case 4a (Johansen-Hendry-Juselius).’ This is how “Table 3: Panel Johansen–Hendry–Juselius cointegration test results” was generated. 6. For impulse response functions, use view\Impulse responses menu, for Responses select only ‘sovereign_premium’, set horizon length to 20, for standard errors and confidence intervals the method for SEs and Cis is ‘Bootstrap: Hall’s studentized’, with ‘display intervals using lines’ and in simulation options tick in the ‘Fast double bootstrap approximation’ box. This is how “Figure 2: Impulse response of sovereign spread to the selected variables (2008Q1–2019Q4)” and “Figure 3: Impulse response of sovereign spread to the selected variables (2008Q1–2022Q3)” were created. 7. For variance decompositions, use view\variance decompositions menu with default settings, this is how “Table 4: Variance decomposition of sovereign spread (2008Q1–2019Q4)” and “Table 5: Variance decomposition of sovereign spread (2008Q1–2022Q3)” were estimated. 8. For “Appendix 1 Descriptive statistics of the differentiated variables” please select all the endogenous variables in the workfile and open a group, where view\descriptive stats menu with ‘Common sample’ settings will provide descriptive statistics, while for unit-root tests, you have to check each variable individually with view\unit root test.