《IMF-新冠肺炎疫情爆发后新兴市场的主权银行关系（英）-2022.11-54正式版.docx》由会员分享，可在线阅读，更多相关《IMF-新冠肺炎疫情爆发后新兴市场的主权银行关系（英）-2022.11-54正式版.docx（63页珍藏版）》请在taowenge.com淘文阁网|工程机械CAD图纸|机械工程制图|CAD装配图下载|SolidWorks_CaTia_CAD_UG_PROE_设计图分享下载上搜索。
1、WORK-NG PAPERINTERNATIONAL MONETARY FUNDThe Sovereign-Bank Nexus in Emerging Markets in the Wake of the COVID-19 PandemicPrepared by Andrea Deghi, Salih Fendoglu, Tara Iyer, Hamid Reza Tabarraei, Yizhi Xu and Mustafa Y. YeniceWP/22/223IMF Working Papers describe research in progress by the author(s)
2、 and are published to elicit comments and to encourage debate.The views expressed in IMF Working Papers are those of the author(s) and do not necessarily represent the views of the IMF, its Executive Board, or IMF management.2022NOVtaking. In addition, we extend previous studies on the real effects
3、of sovereign -driven corporate downgrades by shedding light on their potential spillover effects on banks asset quality. All of these represent important contributions to the empirical literature on the sovereign -bank nexus, which improve the understanding of the channels of risk transmission for e
4、merging market economies. Details on the sample, identification strategies and additional empirical analysis are in the Annexes l-IILBl. Developments in the Sovereign-Bank Nexus in Emerging Markets: Some Stylized FactsThe average public-debt-to-GDP ratio in emerging markets surged to a record 67 per
5、cent in 2021 from about 52 percent before the pandemic, as economic activity declined and governments greatly increased fiscal support to nonfinancial firms and households to cushion the impact of the crisis (Figure 2, panel 1). Although public debt levels have also risen in advanced economies, the
6、domestic sovereign debt exposure of banks has increased relatively more in emerging markets (Figure 2, panel 2)一reaching 17 percent of total banking sector assets in 2021as the additional government financing needs have been met mostly by domestic banks amid declining foreign participation in local
7、currency bond markets (Figure 2, panel 3).The overreliance of governments on domestic banks for their financing needs amid a higher exposure of banks to sovereign debt increases, in turn, the likelihood of shock transmission between the two sectors. This is reflected, for instance, in the positive c
8、orrelation between sovereign and bank expected default frequency, a proxy of default risk (Figure 2, panel 4). The relationship is not only persistent across time, but it is also much tighter when global financial conditions are under strain, such as after the onset of the pandemic in March 2020. Th
9、e time-varying correlation coefficient between sovereign and bank default risk is computed with a 24 -month rolling window. The relationship jumped during the global financial crisis and at the onset of the COVID-19 pandemic in March 2020 to about 0.6. Notably, the strength of this relationship vari
10、es with the level of distress in the banking sector: at low levels of bank dis tress. In unreported results, we find that a 1 percentage point increase in sovereign default risk is associated with a 0.4 percentage point increase in banks* expected default frequency. However, the association is 10 ti
11、mes stronger at higher levels of distress.Banks in emerging markets are generally better capitalized than in the past because of reforms enacted following the global financial crisis and the policy support provided during the pandemic. Average capital adequacy ratio in 2021 was thus 18 percent in 20
12、21 up 3 percent from the average in 2008 (April 2022 GFSR). However, sovereign debt exposure constitutes a significant share of regulatory capital in some countries.Furthermore, a sizable share of banks, sovereign debt holdings follows mark-to-market accounting in several emerging markets, which cou
13、ld undermine banks5 capital adequacy if the market value of these assets were to decline. In some major emerging markets, banks hold floating -rate bonds, inflation-indexed bonds, and Unon-defaultablen bills issued by central banks, which may be less sensitive to interest rates and sovereign risk an
14、d could provide some insulation from a rise in sovereign risk. Sovereign stress could thus potentially quickly transmit to the banking sector in these economies. In this regard, it is worth noting that countries whose banks are more exposed to sovereign debt are also those with a higher public debt-
15、to-GDP ratio and lower bank capital ratios (Figure 2, panels 5 and 6).Figure 2. Developments in Emerging Market Public Debt and Banks9 Sovereign Exposures1. Public Debt, 2005-21 (Percent of GDP)2. Banks Domestic Sovereign Debt Exposure, 2005-21Sovereign-banksBanks-NF Cs Sovereign-NF Cs Glob al fi na
16、n cial co ndi tio ns (righ t seal e)Change in Local Currency Sovereign Bond Holdings (Billions of US dollars, cumulative change since end-2019)Jan. Apr. 20 Jul. 20 Oct. 20 Jan. 21 Apr. 21 Jul. 21 Oct. 21 2020Public Debt and Banks Holdings of SovereignDebt, 2021 (Percent)Median Correlation among Sove
17、reign, Banks and Nonfinancial Corporate Sector Stress, and Global Financial Conditions, 2008:Ml-2021:M9 (Index) Tier 1 Capital-to-Total-Assets Ratio and Banks13. Median Correlation among Sovereign, Banks and Nonfinancial Corporate Sector Stress, and Global Financial Conditions, 2008:Ml-2021:M9 (Inde
18、x) Tier 1 Capital-to-Total-Assets Ratio and Banks1 Holdings of Sovereign Debt, 2021 (Percent)Ratio of public debt lo GDPRatio of Tier 1 capital to total assetsSources: Fitch Connect; IMF, Financial Soundness Indicators, Monetary and Financial Statistics, World Economic Outlook, and Fiscal Monitor da
19、tabases; and authors* calculations.Note: In panels 1 and 2, indicators are country averages weighted by purchasing-power-parity GDP. Public debt is in real terms; that is, in trillions of chained 2010 US dollars. In panel 2, banks* sovereign exposure corresponds to claims on central government debt
20、divided by total banking sector assets. Emerging markets are identified according to the IMFs Vulnerability Exercise for Emerging Market Economies classification. Advanced economies comprise economies classified as advanced in the IMF World Economic Outlook database. Pane I 4 shows the median time-v
21、arying correlation between changes in sovereign, bank, and nonfinancial corporation EDFs across countries using a 24- month rolling window. The median correlation is a number between -1 and 1. The global financial conditions indicator refers to the common component of monthly equity price returns es
22、timated across advanced economies and emerging markets using a factor- augmented vector autoregressive model. In panel 5, red dots reflect public-debt-to-GDP ratios in 2021 vis-a-vis banks central government debt holdings in 2021 (third quarter). In panel 6, total assets are used in the denominator
23、of the Tier 1 capital ratio (instead of risk-weighted assets) to provide greater comparability across countries. Given limited country-level data availability, banks, sovereign debt exposures for India and Argentina are computed using bank-level Fitch Connect data. Data labels use International Orga
24、nization for Standardization (ISO) country codes. AEs = advanced economies; EMs = emerging markets; NFCs = nonfinancial corporations.It is worth noting that banking sector health also depends on the viability of banks1 corporate borrowers, which have faced strains during the pandemic. In most emergi
25、ng markets, the sustainability of corporate debt-as measured by earning capacity relative to debthas declined as corporate revenues have fallen. While it is still difficult to fully ascertain the soundness of bank balance sheets at the current juncture because of regulatory flexibility and other fin
26、ancial sector support measures in place, nonperforming loans are more than one-tenth of total loans in some countries and could edge up as policy support measures are unwound to curb inflationary pressures and financial conditions tighten (April 2022 GFSR).IV. Evaluating the Strength of the NexusTo
27、quantify the overall strength of the nexus in emerging markets, two-way relationships between the sovereign, banking, and nonfinancial corporate (NFC) sectors default risks are examined for individual countries, while controlling for relevant domestic and external factors that may impact these relat
28、ionships. A o structural vector autoregression model (SVAR) is estimated at the country-level using daily data. The SVAR model takes the following form:= + | -1 + o + + - (1)where t indicates time, is a vector of endogenous variables capturing sovereign default risk, bank default risk, non-financjal
29、 corporate default risk, term spread and equity indices. Default risk is proxied by expected default frequency (EDF). with = 1, ., and r with = 1, ., , are coefficient matrices. The matrix contains the contemporaneous effects of structural shocks on the endogenous variables and allows to track the s
30、trength of linkages from one sector to another. is the vector of exogenous variables, including a measure of global financial conditions (or U.S. monetary policy shocks) and the daily return on a trade -weighted dollar index. See Annex I for more details and description for this and following sectio
31、ns.where t indicates time, is a vector of endogenous variables capturing sovereign default risk, bank default risk, non-financjal corporate default risk, term spread and equity indices. Default risk is proxied by expected default frequency (EDF). with = 1, ., and r with = 1, ., , are coefficient mat
32、rices. The matrix contains the contemporaneous effects of structural shocks on the endogenous variables and allows to track the strength of linkages from one sector to another. is the vector of exogenous variables, including a measure of global financial conditions (or U.S. monetary policy shocks) a
33、nd the daily return on a trade -weighted dollar index. See Annex I for more details and description for this and following sections. The trade-weighted US dollar index, also known as the broad index, is a weighted average of the foreign exchange value of the US dollar against the currencies of a bro
34、ad group of major U.S. trading partners. To account for non-stationarity of the EDF data, the model is estimated in first differences for the three EDF variables. For equity indices, log differences are used. The term is a vector of structural shocks with diagonal variance matrix = ().To identify st
35、ructural shocks to the endogenous variables, the analysis exploits the heteroskedasticity in the data following Rigobon (2OO3). Identification through heteroskedasticity yields consistent estimates even if the regimes are misspecified (see Rigobon, 2003). Alternatively, a structural form-GARCH can b
36、e used. One challenge related to the identification strategy is the interpretation of the structural shocks that are not based on priori assumptions. Although more restrictive, alternative standard identification strategies could be used for robustness. For instance, zero restrictions could be impos
37、ed on A, arguing for delayed responses of some endogenous variables to the others. Sign restrictions on A could also be used allowing for contemporaneous effects among all variables. This identification strategy relies on the fact that changes in the volatility of structural shocks contain additiona
38、l information on the relationship between the endogenous variables. Thus, large shocks to bank risk, for example, are considered to contain more information about the response of sovereign risk to bank risk as the covariance between the two types of risks increases. Thus, bank risk shocks can be use
39、d as a probabilistic instrumenf to trace out the response of sovereign risk.1. Effect of a One Standard Deviation Shock on Other Sectors Default Risk (Standard deviation)0.3Estimated range Average effectSovBanksBanksNFCBaBanSov s 一nkksNFCsSovNFCsSovsCsSources: Haver Analytics; Moodys; Refinitiv Data
40、stream; and IMF staff calculations.Figure 3. Transmission of Risks through the Sovereign-Bank Nexus: Strength of the Main Channels across Emerging MarketsCH NNote: The figure shows the estimated range of coefficients for 11 individual emerging markets obtained from a structural mode I using daily da
41、ta of default risk for sovereign, banking, and corporate sectors. NFCs = nonfinancial corporations; Sov = sovereign .Three key findings emerge from this analysis, which are presented in Figure 3. First, the nexus is strong, with significant spillover effects between sectors (Figure 3, panel 1). The
42、estimated spillover effects are statistically and economically significant in most of the countries of the sample. A standard deviation increase in sovereign risk increases bank EDF directly by 5 percent of its standard deviation on average, while the same shock increases NFCs risk by 6 percent stan
43、dard deviation.Second, the strength of the transmission of risk between sectors varies. For example, spillovers from sovereign default risk to banks are, on average, larger than those in the opposite direction from banks to sovereign default risk. Overall, the largest spillovers are from sovereign a
44、nd bank to firms default risk.Third, the relevance of the nexus differs across countries (Figure 3, panel 2). Across the eleven countries in the sample, the strongest shock transmission from sovereign to banks is found in Col ombia and China, which could reflect a greater presence of state-owned ban
45、ks, or a larger share of marked-to-market sovereign exposures on banks balance sheets. Our results are similar to the findings of Acharya, Drechsler, and Schnabl (2014) based on data released as part of the 2010 Eurozone bank stress tests. These authors1 estimates imply that a 1 standard deviation i
46、ncrease in sovereign CDS spread translates into about 5 percent increase in bank CDS spreads1 standard deviation. Our specification using high frequency data for the period from 2007 to 2020 finds that the transmission of shocks from sovereign to banks have a similar average effect but can reach on
47、average about 15 percent of banks default risk standard deviation for economies with the strongest spillover effects. Data compiled from banks accounting statements and Basel Pillar III disclosures for Colombia, indicate a larger share of mark - to-market sovereign bonds reaching almost 80 percent o
48、f the total sovereign exposures in 2020. For instance, a one standard deviation shock to the sovereign EDF increases banks EDF by about 0.15 standard deviation. Together, the estimates suggest a significantly positive two -way causality between sovereign and bank risk that could generate a feedback
49、loop.The heterogeneity in the size of the transmission of shocks suggests that some country -specific factors, such as the fiscal position and financial vulnerabilities, may be at play in amplifying the impact of an adverse shock.Further empirical analysis supports this observation. For example, after a sharp tightening in global financial conditions, emerging