A. Introduction

1. Rising sovereign debt in the wake of the COVID-19 pandemic has generated renewed attention to the financial sector-sovereign nexus in South Africa. In the absence of fiscal space, the necessary measures to support the economy following the pandemic prompted a further increase in the stock of sovereign debt and worsened indicators of sovereign risk (Box 1 and Figure 1). The higher sovereign debt, in turn, tightened the nexus between the sovereign and the domestic financial system, notably banks, pension funds, insurance companies, and mutual funds in the context of reduced purchases and greater disposals of sovereign debt by nonresident investors. In addition to the direct channel (public debt acquisition by the financial sector), the nexus has indirect channels, including an exposure of financial institutions to domestic economic activity.

Indicators of Sovereign Risk

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, Morgan Markets, and IMF staff calculations.Note: Broken lines are + /-one standard deviation around the mean using 2010–19 data.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Indicators of Sovereign Risk

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, Morgan Markets, and IMF staff calculations.Note: Broken lines are + /-one standard deviation around the mean using 2010–19 data.

• Download Figure
• Download figure as PowerPoint slide

Indicators of Sovereign Risk

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, Morgan Markets, and IMF staff calculations.Note: Broken lines are + /-one standard deviation around the mean using 2010–19 data.

• Download Figure
• Download figure as PowerPoint slide

2. Rapid increases in the financial sector-sovereign nexus have been a global matter of concern, particularly since the late-2000s. During the Global Financial Crisis in the late-2000s, the public debt-to-GDP ratio rose across many countries, especially in the European periphery. Prompted by foreign investors’ flight and cheap ECB funding, many peripheral European banks deepened their home bias, absorbing sizeable amounts of domestic sovereign debt, both in the primary and secondary markets. As the bank-sovereign nexus became entrenched, concerns regarding the health of the banking sector rose. These concerns were due, inter alia, to rising asset quality problems, including valuation losses on banks’ sovereign debt holdings on the one hand, and sovereign credit quality as governments provided guarantees or other support to their banking systems, on the other. An increased bank-sovereign nexus was also observed across many advanced economies. In this regard, the IMF (2015) examined the complex linkages between the health of the banking system and sovereign debt, including implications for fiscal and monetary policy, while Dell’Ariccia et al. (2018) provided a broad overview of ways of managing the bank-sovereign nexus.

3. This paper documents several aspects of the financial sector-sovereign nexus in South Africa and policy discussions about how to mitigate associated risks. It examines the size of sovereign debt in financial intermediaries’ balance sheets, measures the local currency sovereign risk, and discusses the transmission of risks from the government to the financial sector and vice versa. It surveys the literature on the fiscal cost of banking crises—strong linkages between banks and the sovereign could substantially weaken bank’s balance sheets, and government interventions have been found to be expensive in cases of banking problems in some countries. This nexus is likely to remain important, and the paper surveys some recommendations, in line with the 2021 Financial Stability Assessment Program (FSAP), on how to limit risks from the bank-sovereign nexus.

B. Importance of Sovereign Debt for Financial Intermediaries’ Balance Sheets

4. Globally, banks hold domestic sovereign debt for a number of important reasons.² International prudential standards set by the Basel Committee on Banking Supervision (Basel standards) provide national discretion in treating bank holdings of domestic sovereign debt with respect to risk weights, large exposures, market risk, and credit risk mitigation, thus leaving room for regulatory incentives (BCBS, 2017; Dell’Ariccia et al., 2018). The preferential treatment given to sovereign debt relative to other financial assets in domestic regulatory frameworks is likely amplified during economic downturns (IMF, 2015). Sovereign debtis considered as a safe and high-quality asset for banks to meet the liquidity requirements, a strong collateral asset for central bank operations and secured wholesale funding, and a benchmark for pricing financial assets. Sovereign debt could also represent an important source of income particularly when income from other sources underper form.

5. In addition to the incentives mentioned above, country authorities often take policy actions to further support banks’ holdings of government debt during times of stress. As discussed by Asonuma, Bakhache, and Hesse (2015a), these actions could include liquidity extension to banks, direct purchases of government debt, and/or conditional commitments to purchase government debt by central banks. Financial repression and moral suasion are sometimes used to ‘convince’ banks to purchase government bonds, especially in the primary market.³ At the same time, the supply of public debt often substantially increases during times of stress, including as a result of countercyclical fiscal policy. With the quality of other assets deteriorating, domestic banks tend to prefer holding sovereign debt to help safeguard the health of their balance sheets. In addition, private-sector investment opportunities tend to decline during times of stress, further pushing banks toward domestic sovereign debt holdings.

6. Several factors have been identified as important drivers of bank holdings of sovereign debt across countries. Using a sample of advanced and emerging market economies (EMEs), Asonuma, Bakhache, and Hesse (2015a) show that banks’ bias to invest in domestic sovereign debtis associated with high uncertainty and increasing inflation (potentially capturing signs of macroeconomic instability or increased moral suasion). In contrast, the private-sector credit-to-GDP ratio (partly reflecting banks’investment opportunities outside the government) and institutional quality (capturing government stability and socioeconomic conditions) are significantly negatively related to home bias. Moreover, Dell’Ariccia etal. (2018) provide empirical evidence that banks hold more government debt during periods of high interest rates and in countries with lower private-sector credit-to-GDP ratios. Banks operating in less developed financial systems—for instance, with fewer high-quality lending opportunities—also hold more government debt.

7. High sovereign debt holdings by financial institutions could create important problems. For instance, high bank holdings of domestic sovereign debt (home bias) may be associated with low private-sector credit growth in emerging and developing countries, mainly reflecting a portfolio rebalancing of banks toward safer and more liquid assets in times of stress (Bouis, 2019). Issues surrounding banks’ home bias in sovereign debt holdings especially came to the forefront during the Euro area crisis. Studies on eurozone countries highlight that, inter alia, fiscal space, changes in perceived sovereign credit quality, and state ownership of banks contribute to the increased propensity of banks to hold domestic sovereign debt.⁴

8. Many of these factors are likely to remain relevant for South Africa. High-quality liquid assets (HQLA) eligibility of sovereign debt, amid shortages of other HQLA eligible assets, and phasing out of the SARB’s committed liquidity facility (CLF) provide banks with incentives to hold sovereign debt.⁵ The largest six banks, on average, hold close to 90 percent of HQLA in Level 1 unencumbered assets, which are mainly in domestic government securities and central bank reserves. The largest banks act as primary dealers in the sovereign debt market, absorbing and passing on the debt, and market makers in the secondary market (SARB, 2021). Risks from the nexus to both the financial system and the sovereign will increase absent sufficient fiscal consolidation to keep the supply of sovereign debt in check. Risks will also increase if prospects for private investment, demand for bank credit, and broader economic activity remain weak, all of which are also partly constrained by remaining structural rigidities and limited options to diversify away for HQLA purposes.

C. Government Bond Holdings by Banks and Nonbank Financial Institutions in South Africa

9. Holdings of government securities relative to assets have increased for banks but remained relatively low for nonbanks. An analysis of the largest 6 banks (“top 6”), representing a little over 90 percent of system assets, and the rest (“other banks”) shows that the holdings of government securities, bonds, and T-bills by the top 6 moderated from 6–8 percent of assets in the 1990s to nearly 4 percent in early-2008. Since then, such holdings rebounded to around 8 percent of assets by end-2011, with bonds representing around 60 percent. Bank holdings of government securities started to rise again in 2017 and reached somewhat above 12 percent of assets by January 2021. The increase was driven mainly by bond holdings, raising the bonds’ share of total holdings to 75 percent. Holdings of government securities by “other banks” marginally moderated to somewhat below 5 percent of assets by early-2008. Since then, such holdings rose to nearly 25 percent of assets in January 2021. The increase was due mainly to T-bill holdings, taking their share of total securities holdings to around 70 percent. By contrast, nonbank financial institutions shed their government securities holdings from close to 35 percent of assets in the 1990s to around 15 percent in the 2000s. The ratio rose moderately to 17 percent by mid-2020, due mainly to holdings by “other financial institutions” (OFIs), which include mutual funds and collective investment schemes.

Bank and Nonbank Financial Institutions’ Holdings of Government Securities

(Percent of respective system assets, stacked)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, IMF IFS, and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Bank and Nonbank Financial Institutions’ Holdings of Government Securities

(Percent of respective system assets, stacked)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, IMF IFS, and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Bank and Nonbank Financial Institutions’ Holdings of Government Securities

(Percent of respective system assets, stacked)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, IMF IFS, and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

10. Cross-country comparisons suggest that relative to assets, government bond holdings by the South African financial sector are broadly comparable to the “norms”. Data for South Africa are compared to those for all countries for which data are available. South African banks’ government bond holdings as a share of assets were relatively low in the early-2000s, near the first quartile of 99 countries. Holdings then moved up to the cross-country median in the late-2000s, and broadly tracked the gradual increase in the median to date. Unlike banks, nonbanks held relatively large amounts of government bonds as a share of assets in the early-2000s—at around the third quartile of 37 countries. After declining toward the cross-country median in the early-2000s, the ratio remained broadly flat to date, thus closing the distance from the cross-country median, which gradually increased. As of September 2020, South African bank holdings of government bonds were somewhat below the cross-country median of 129 countries, but above the levels in Brazil, Mexico, and Turkey. South African nonbanks were positioned somewhat below the cross-country median of 62 countries, similar to Turkey and above Mexico.

Bank and Nonbank Financial Institutions’ Holdings of Government Securities

(Percent of assets)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, IMF IFS, and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Bank and Nonbank Financial Institutions’ Holdings of Government Securities

(Percent of assets)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, IMF IFS, and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Bank and Nonbank Financial Institutions’ Holdings of Government Securities

(Percent of assets)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, IMF IFS, and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

11. Relative to the stock of bonds outstanding, holdings by nonbanks and nonresidents appear relatively high in South Africa. Compared with selected EMEs (in the Sovereign Debt Investor Base), South African banks have notably increased their government bond holdings as a share of the total stock during the COVID-19 pandemic, but their holdings still remain below the EME median. Nonbanks progressively reduced their government bond holdings as a share of the total stock through the early-2010s, but still remain in the top quartile. Nonresident holdings rose from around the median in the early-2000s to the top quartile in the mid-2010s and have remained there even after nonresidents sold government bonds during the COVID-19 pandemic.

12. Looking ahead, the cost of funding for the sovereign, and for the economy more broadly, could increase. A simple correlation analysis suggests that banks in South Africa tend to increase their holdings of government securities when the yield curve steepens (potentially as government securities’ valuation becomes more attractive). Moreover, a recent SARB econometric analysis (Makrelov et al., 2021) suggests that higher fiscal risks would prompt banks to increase their capital as a mitigant, making it more expensive to hold such bonds.⁶ Indeed, IRB banks have been increasing risk weights for sovereign exposure up to 10 percent in line with the rising public debt ratio and weakening sovereign credit ratings. In addition, South African sovereign bond valuations could become more volatile, potentially increasing the risk of large valuation losses. Such losses could compress profitability and capitalization to the extent that the larger four banks mark to market roughly one half of their government bond holdings. Results from the FSAP’s stress tests warn about the vulnerability of banks to a weakening of sovereign credit quality.⁷ As a result, banks would demand higher yields to hold sovereign debt.

Resident and Nonresident Holdings of Emerging Market Local Government Bonds

(Percent of stock of government bonds)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, IMF Sovereign Debt Investor Base, and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Resident and Nonresident Holdings of Emerging Market Local Government Bonds

(Percent of stock of government bonds)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, IMF Sovereign Debt Investor Base, and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Resident and Nonresident Holdings of Emerging Market Local Government Bonds

(Percent of stock of government bonds)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, IMF Sovereign Debt Investor Base, and IMF staff calculations.

• Download Figure
• Download figure as PowerPoint slide

Term Premia and Bank Holdings of Government Securities

(Bank holdings of government securities, percent of total assets)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, SARB BA 900, and IMF staff calculations.Note: Top 6 is separated into Top 4 and Mid 2.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Term Premia and Bank Holdings of Government Securities

(Bank holdings of government securities, percent of total assets)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, SARB BA 900, and IMF staff calculations.Note: Top 6 is separated into Top 4 and Mid 2.

• Download Figure
• Download figure as PowerPoint slide

Term Premia and Bank Holdings of Government Securities

(Bank holdings of government securities, percent of total assets)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: Haver, SARB BA 900, and IMF staff calculations.Note: Top 6 is separated into Top 4 and Mid 2.

• Download Figure
• Download figure as PowerPoint slide

^{Table 1.}

Accounting Treatment of Bank Holdings of Government Bonds

(Percent of total)

Sources: Fitch connect and IMF staff calculations.

^{Table 1.}

Accounting Treatment of Bank Holdings of Government Bonds

(Percent of total)

Bank	Trading / FV thr. Income	Available for Sale	Held to Maturity
1	67.6	0.4	32.0
2	53.5	02	46.3
3	52.1	26.9	21.1
4	44.3	3.2	52.5

Sources: Fitch connect and IMF staff calculations.

View Table

^{Table 1.}

Accounting Treatment of Bank Holdings of Government Bonds

(Percent of total)

Bank	Trading / FV thr. Income	Available for Sale	Held to Maturity
1	67.6	0.4	32.0
2	53.5	02	46.3
3	52.1	26.9	21.1
4	44.3	3.2	52.5

Sources: Fitch connect and IMF staff calculations.

D. Fiscal Risks’ Spillovers and Feedback Loops

13. There are several key channels through which sovereigns could affect banks. Dell’Ariccia et al. (2018) and SARB (2021) identify exposure, safety nets, and macroeconomic linkages as key channels. The exposure channel reveals that bank holdings of sovereign bonds are adversely affected by falling sovereign bond prices, which could also lead to higher bank wholesale funding costs as the collateral value of sovereign bonds falls.⁸ The safety net channel refers to the contingent liabilities that governments incur given their traditional role as main backstops in case of banking problems, which could result in market concerns if fiscal space is limited, feeding into lower bank profits.⁹ With sovereign and bank ratings inherently intertwined, sovereign rating downgrades are passed on to bank ratings, usually increasing bank funding costs. There are also pertinent macroeconomic [inks that can propagate shocks from higher public sector deficits and debt to higher sovereign and domestic interest rates, adversely impacting bank balance sheets.

14. There are channels of spillbacks through which the banking sector can impact the public sector balance sheet directly and indirectly. IMF (2015) argues that the direct effect occurs when the government intervenes in the banking sector to manage a crisis (contingent liabilities become real liabilities and worsen the debt outlook) and the indirect one occurs when banking-sector developments affect the main drivers of debt (growth, primary balance, and interest rate). IMF (2015) shows that the more an economic boom is driven by banks, the deeper is the ensuing recession, with a longer recovery compared to a boom-bust cycle driven by nonbanks. Similarly, the fiscal sector’s “boom-bust” cycle is more pronounced and damaging when it is driven by the banking sector.

15. In addition, banks’ home bias in their sovereign debt holdings tends to delay fiscal consolidation until debt reaches dangerously high levels. The propensity of banks to hold domestic sovereign debt over foreign sovereign debt creates a captive investor base and may provide greater fiscal breathing space, potentially delaying the necessary fiscal adjustment. Drawing on an estimation of fiscal reaction functions for advanced and emerging economies, Asonuma, Bakhache, and Hesse (2015b) show that when banks exhibit higher home bias in their sovereign debt holdings, fiscal consolidation by the sovereign tends to be slower, ceteris paribus.¹⁰ According to Figure 6, a relatively high banking sector home bias (relative to the sample median) is associated with a substantially weaker average primary fiscal balance (relative to the sample of economies with a given level of public debt, here 80 percent of GDP, somewhat above the 70 percent of GDP in 2020 for South Africa). Dell’Ariccia et al. (2018) and Ongena et al. (2016) find evidence of moral suasion during the Euro area crisis, where domestic banks play a greater role than foreign banks in accommodating higher sovereign financing needs.

Impact of Home Bias on Primary Balance

(Percent of GDP, when debt is 80 percent of GDP)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Source: Asonuma, Bakhache and Hesse (2015b).Note: Home bias (HB) is defined as banks’ holding of domestic sovereign claims in total assets. Low (high) HB denotes the average of observations whose HB is below (above) the median in the estimation of the fiscal reaction function.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Impact of Home Bias on Primary Balance

(Percent of GDP, when debt is 80 percent of GDP)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Source: Asonuma, Bakhache and Hesse (2015b).Note: Home bias (HB) is defined as banks’ holding of domestic sovereign claims in total assets. Low (high) HB denotes the average of observations whose HB is below (above) the median in the estimation of the fiscal reaction function.

• Download Figure
• Download figure as PowerPoint slide

Impact of Home Bias on Primary Balance

(Percent of GDP, when debt is 80 percent of GDP)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Source: Asonuma, Bakhache and Hesse (2015b).Note: Home bias (HB) is defined as banks’ holding of domestic sovereign claims in total assets. Low (high) HB denotes the average of observations whose HB is below (above) the median in the estimation of the fiscal reaction function.

• Download Figure
• Download figure as PowerPoint slide

16. In South Africa, the increasing interrelationship between banks and the government poses challenges going forward. All three channels (exposure, macroeconomic, and safety net linkages) are relevant for South Africa.

• Exposure channel: Banks hold government bonds for the number of reasons discussed earlier. The valuation losses on sovereign bond holdings (and the collateral used for funding) could pressure banks’ profitability, capital position, and funding.

• Macroeconomic channel: Both banks and the sovereign affect and are affected by macroeconomic aggregates, such as output growth, fiscal policy, and interest rates. Since 2010, all successive foreign currency sovereign downgrades by the main three rating agencies to eventually below investment grade (6 downgrades in total) have been matched by downgrades of banks’ credit ratings, as the latter are capped by the sovereign ratings. With the fiscal deficit projected to remain elevated amid steadily increasing public debt levels, the absorptive capacity of the banking system will be key if nonresidents take a cautious stance.

• Safety net channel: Moral suasion and financial repression might have a short-term benefit in terms of having a captive domestic investor base and providing additional fiscal breathing space. However, this comes at the potentially high cost of making the bank-sovereign nexus even more vulnerable to shocks. The government tends to provide a backstop to banks, which creates a linkage between the credit quality of banks and the sovereign. Perception of a less credible backstop would weaken the financial sector’s perceived credit risk and lead to higher funding costs than otherwise. In turn, perception of a higher chance (and scale) of fiscal support would reduce the sovereign’s perceived credit quality. More broadly, it is important, inter alia, to finalize the Financial Sector Laws Amendment Bill (FSLAB) to improve the bank resolution framework and introduce a deposit guarantee scheme.

^{Box 2.}

Feedback of Bank and Sovereign Risks

The correlation between bank credit risk and sovereign credit risk has increased. Daily CDS spreads estimated from the expected default frequency (EDF) are used to calculate the two-year moving correlation between the sovereign and ten banks for which the indicator is available from Credit Edge. The correlation has increased, and the dispersion has tightened over the years, mainly as small banks’ correlation with the sovereign moved from either negative or small positives to levels comparable with large banks’ correlation. Some of the increases likely reflect perception of higher risk, which tends to push the asset price correlation higher.

Indicator of Bank-Sovereign Nexus

(Correlation coefficients, median and quartiles around)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: CreditEdge and IMF staff calculations.Note: Based on 2-year rolling correlation FVCDS levels for 10 banks and the sovereign. FVDS = credit default swap (CDS) spreads implied by expected default probability (EDF).

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Indicator of Bank-Sovereign Nexus

(Correlation coefficients, median and quartiles around)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: CreditEdge and IMF staff calculations.Note: Based on 2-year rolling correlation FVCDS levels for 10 banks and the sovereign. FVDS = credit default swap (CDS) spreads implied by expected default probability (EDF).

• Download Figure
• Download figure as PowerPoint slide

Indicator of Bank-Sovereign Nexus

(Correlation coefficients, median and quartiles around)

Citation: IMF Staff Country Reports 2022, 038; 10.5089/9798400201318.002.A001

Sources: CreditEdge and IMF staff calculations.Note: Based on 2-year rolling correlation FVCDS levels for 10 banks and the sovereign. FVDS = credit default swap (CDS) spreads implied by expected default probability (EDF).

• Download Figure
• Download figure as PowerPoint slide

E. Fiscal Cost of Banking Crises in the Literature

17. Over the past four decades, banking crises have contributed to large output losses and fiscal costs. Empirical evidence shows that the median output losses from banking crises are 35 percent of GDP in high-income countries and 14 percent of GDP in low- and middle-income countries. In the former, the larger size of their financial systems and longer crisis duration contributed to the higher output cost(Laeven and Valencia, 2018). Similarly, the median increase in public debt in the four years after a banking crisis is larger in high-income countries (a little over 20 percent of GDP) than in low- and middle-income countries (16–17 percent of GDP), probably reflecting larger fiscal space in high-income countries allowing them to pursue greater countercyclical policies and use automatic stabilizers (IMF, 2015). There is also a large variation across countries— during the Asian financial crisis, Indonesia’s fiscal costs reached more than 50 percent of GDP, while during the global financial crisis, fiscal costs in Iceland and Ireland exceeded 30 percent of GDP (Dell’Ariccia et al., 2018).

18. A number of factors drive the fiscal costs of banking crises. Direct fiscal costs of banking crises were higher in countries where banks were more leveraged and reliant on external wholesale funding prior to the crises (IMF, 2015). Banks that are dependent on external wholesale funding usually face higher rollover risks and possibly solvency risks that may necessitate greater public funds for preemptive recapitalization. Countries that guarantee the entire bank liabilities during a crisis may limit up-front deposit disbursements or issuance of debt, but face, on average, higher direct fiscal costs (see also Dell’Ariccia et al., 2018). By contrast, direct fiscal costs are found to be lower in countries with higher quality of supervision and greater credibility that the government would provide sufficient financial safety nets, such as broad deposit insurance coverage. Also, swifter government intervention tends to lower fiscal costs ex post (Laeven and Valencia, 2010).

19. There are several lessons for South Africa from the experiences of banking sector stress elsewhere. In general, South Africa’s banks have strong capital and liquidity buffers and are well regulated. They have navigated both the Global Financial Crisis in the late-2000s and the ongoing pandemic well. Stress tests by the joint IMF-World Bank FSAP show their resilience to adverse shocks. However, rising public-sector fiscal deficits and debt amid an increase in sovereign risk premia could lead to concerns about banks’ increasing holdings of domestic sovereign debt. As another important channel, any prolonged macroeconomic difficulties could be challenging for the domestic banking system despite their buffers and good management.

F. Mitigating the Financial Sector-Sovereign Nexus Risks

20. Macroeconomic policies are the first line of defense and using potential regulatory measures would take the authorities into new territory. Fiscal consolidation and structural reforms that IMF staff has recommended will help reduce the supply of sovereign debt, boost medium-term growth, improve banks’ lending opportunities, and further strengthen bank capital buffers. Mitigating risks from the financial sector-sovereign nexus using regulatory measures would be a new approach, with few countries currently choosing to resort to such measures.

21. The literature has advocated several key principles as to how risks associated with the financial sector-sovereign nexus could be addressed. Finalizing the FSLAB, enhancing the resolution framework, and introducing a deposit guarantee scheme will go a long way in enhancing the safety net channel. Dell’Ariccia et al. (2018) argue that buffers would usefully reduce risks from the bank-sovereign nexus—stronger bank capital, sound fiscal positions, and prudent macro-structural policies help reduce risks. Avoiding measures, such as onerous concentration limits, that might have unintended pro-cyclical consequences, is important. Unintended consequences could take the form of an excessive reduction of liquidity, bond market pressures, or other unwarranted macro-financial dynamics. These unintended consequences could be especially problematic during a sharp economic downturn amid declining sovereign bond prices.

22. The following carefully calibrated regulatory measures to alleviate the bank-sovereign nexus, discussed during the 2021 FSAP mission, will be useful:

• Increasing risk weights on sovereign bond holdings. IRB banks have already increased risk weights on their holdings of domestic sovereign debt and therefore hold more capital against them. Similarly, under national discretion, a (relatively low) risk weight may be applied to domestic sovereign debt denominated and funded in domestic currency while being mindful of potential pro-cyclical effects.

• Applying Pillar 1 or 2 capital surcharges. Surcharges could be applied on holdings of domestic sovereign bonds only above certain thresholds. Such surcharges would be calibrated to reflect perceived risks and discourage excessive concentration, while limiting risks of unintended side-effects (e.g., an overly higher cost of meeting liquidity requirements).

• Introducing a quantitative measure to reduce concentration. As an important downside risk, putting a cap on concentration could create ‘cliff effects’, that is, as bank holdings of domestic sovereign debt suddenly rise either close to or past the limits, banks might quickly shed “excess” bond holdings.

23. The process of potential implementation was also discussed:

• To achieve the objectives, the measures would need to be gradually introduced, carefully calibrated, and clearly communicated.

• The measures could best be introduced after the ongoing normalization of the COVID-19-related prudential requirements has been completed.

• A reasonable transition period will be needed to give banks time to adjust their balance sheets. An announcement of envisaged near-term measures, with the applicable transition period, would help prevent further intensification of the nexus and smooth adjustment.

24. Importantly, regulatory efforts need to be supported by fiscal consolidation to reduce the supply of government debt as well as structural reforms to boost growth durably.