powell20novfig2.png
VoxEU Column COVID-19 Financial Regulation and Banking

Resilience and fragility in global banking: Impacts on emerging economies

Global banks are highly connected, and banking systems are only as strong as the weakest links in the network. This column analyses cross-border syndicated lending to emerging and developing countries from 1993 to 2020 and finds evidence of both resilience and fragility in the global financial system. Contagion through co-lenders affected bank lending more strongly before and during the 2008-09 financial crisis but significantly less in a period after the crisis, consistent with the idea that the reduction in network density as a result of the crisis may have increased resilience to ‘normal shocks’. But Covid-19 is clearly no normal shock, and its impacts are likely spreading through the network, affecting the supply of loans to emerging economies.

While there has been much focus on the resurgence of cases in Europe and the US, emerging economies remain heavily impacted by the Covid-19 pandemic (Financial Times, 2020a).  Five of the eight countries with more than one million cases are emerging economies as of 27 October 2020.1

In the aftermath of the crisis, there were large outflows of capital from emerging economies, and this capital has not fully returned. For example, the withdrawals from bond funds investing in Latin America and the Caribbean amounted to almost 4% of GDP (Nuguer and Powell 2020).

To buffer the economic impacts of the crisis, the public sector response of advanced economies has been unprecedented, with large Covid-19 packages of around 9% of GDP. In emerging economies, fiscal constraints have limited the response and the average package is less than 4% of GDP.2

Central banks have been active through monetary policy and large asset-purchase programmes. But emerging economy central banks also tend to be more constrained. While policy interest rates have fallen and reserve requirements have been reduced, central bank charters in many emerging economies are relatively strict as to what assets can be purchased and how. In emerging economies in Latin America, the memories of the bailouts of both banks and their clients after financial crises in the 1980s, which then fuelled economic instability and inflation, heavily influenced the subsequent design of monetary frameworks, although the Covid-19 crisis has raised the issue as to whether the constraints should be relaxed (IIF 2020). 

Many countries have announced publicly supported guarantee programmes, in which partial guarantees reduce credit risks for banks. However, take-up has been mixed in both advanced and emerging economies as banks appeared reticent to accept the residual risks, other instruments could play an important role especially where uncertainty and leverage is very high (Baldwin and Weder di Mauro 2020, Powell and Rojas-Suarez 2020).

The different measures have reduced the impacts of the crisis on bank balance sheets, but non-performing loans are likely to rise and recovery will be gradual at best amid virus resurgences and as support packages begin to fade (S&P 2020, The Banker 2020, Financial Times 2020b). Ari et al. (2020) review past crises and discuss how higher non-performing loans may be resolved this time around.  Given the weaker public support packages in emerging economies, there are yet greater concerns (Powell and Rojas-Suarez 2020).

Global banks are highly connected and, as the 2008 global financial crisis illustrated, banking systems are only as strong as the weaker links in the network (Allen and Gale 2000). A further concern is the impact of the crisis on banks and hence on the availability of loans to the private sector in emerging economies – another important source of financing.

Interestingly, the academic literature suggests that banking networks might be resilient to some shocks (smaller shocks that hit banks on the fringe of the network) but fragile in the face of others (large shocks to banks that have many connections; the central players). Moreover, the denser the network (the more connections there are between banks) then the higher the stakes. On the one hand, there may be more resilience to those small shocks, as with more connections other banks can step in to take the place of one that has a problem, but if a large shock comes along then this can provoke greater contagion through a denser system.

In a new paper (Conesa et al. 2020), we find evidence of both resilience and fragility in the global financial system. We analyse cross-border syndicated lending to emerging and developing countries from 1993 to 2020. The syndicated lending market is an important source of financing for firms and banks in emerging economies. Historically, international syndicated loans represented some 40 percent of total cross-border flows to emerging markets (De Haas and Van Horen 2012). Syndicated lending is also a bell-weather for debt flows in general as they tend to rise or fall with the overall gross flows of loans from private banks and financial institutions (Figure 1). 

Figure 1 Cross-border syndicated lending is an important element of total gross credit flows to developing countries

Note: The figure shows the co-movement between the nominal amount of Syndicated Loans and the Gross Flow of Loans for low and middle-income countries during 1993-2017. The agreement of new syndicated lending is closely associated with the actual flow (disbursements) of commercial lending to developing countries. Gross flow of loans is the gross flows (disbursements) of non-guaranteed (PNG) long-term commercial bank loans and public and publicly guaranteed (PPG) commercial bank loans from private banks and other financial institutions from World Bank data. 
Source: Authors’ calculations based on Refinitiv and World Bank International Debt Statistics.

Our results indicate that shocks propagate in the cross-border lending network through co-lending relationships, driven mostly by the large global banks who occupy central positions in the network. At the same time, the network is resilient to shocks to banks that are located on the fringes of the network and have limited co-lending relationships. 

Still, the syndicated lending network is not highly centralised or dense. There are some central players with many co-lenders (typically the large global banks) but also many financial institutions on the periphery with relatively few co-lender connections (Figure 2). The network is highly incomplete in the sense that there are few actual connections compared to the many potential ones that could exist. This means that if a borrower is reliant on a loan from a periphery bank that suffers a shock, then it may be hard to find an alternative source for financing.

Figure 2 The lending network in 2017            

Note: Figure 2 illustrates the lending network for 2017. Each bubble represents a bank, and the size of each bubble is in proportion to the number of co-lenders of that institution. The colours show the nationalities of the banks in question. Banks (bubbles) are placed close to each other when they syndicate loans together. As it is common for banks from the same country to form syndicates, bubbles of the same colour tend to be close to each other. At the centre of the network are the large global banks from the United States, Europe and Japan. Taiwan is also important in the cross-border syndicated loan market, and some Chinese banks have now become central players. There are some clusters quite far away from the central mass. Typically, these consist of banks from some emerging countries that also participate in cross-border lending but that do not co-lend much with the central players. 
Source: Authors’ calculations based on Refinitiv.

The global financial crisis severely impacted the network (Figure 3). After 2008-9 the network shrank, with fewer banks (nodes, in network terminology), fewer connections between them (known as edges), and less overall lending. Also, the large global banks became somewhat less central with less co-lender connections and there are new central South-South players, such as the official Chinese banks (Figure 4). 

Figure 3 Network measures: Centrality and density

a) Density            

b) Centrality

Note: This figure shows that average network characteristics are different if we compare the pre and the post GFC periods. In Panel a, Density, which is a measure of how close the network is to complete, has fallen after the GFC. In Panel b, Centrality also decreased in the Post-GFC period, indicating that the banks are less connected to other banks in the network.
Source: Authors’ calculations based on Refinitiv.

Figure 4 Syndicated loan market

Note: The figure shows the evolution of syndicate lending to Emerging and Developing Economies (South) by lenders during 1993-2017. 
Source: Authors’ calculations based on Refinitiv.

Contagion through co-lenders affected bank lending more strongly before and during the global financial crisis, but significantly less in a period after the global financial crisis. This result is consistent with the idea that the reduction in density may have increased resilience to ‘normal shocks’, in accordance with Acemoglu et al. (2015). 

But Covid-19 is clearly no normal shock, and it is affecting many banks including those that are central in the network. Syndicated lending to developing countries in the last months has fallen, as have the number of banks active in the network (Figure 5). While it is too early to assess (and it is a challenge to separate the impact of demand from supply), the impacts are likely spreading through the network affecting the supply of loans to emerging economies. This underlines the need to consider carefully how to assist firms in emerging economies to support employment and livelihoods in the months ahead. 

Figure 5 Syndicated lending to developing economies after the Covid-19 shock by type of finance

Note: The figure illustrates 4-month moving averages of syndicated lending (real USD billions) to developing economies in January 2018-September 2020, both cross-border and national, by announcement date. 
Source: Authors’ calculations based on Refinitiv.

References

Acemoglu, D, A Ozdaglar and A Tahbaz-Salehi (2015), “Systemic Risk and Stability in Financial Networks”, American Economic Review 105(2): 564–608. 

Allen, F and D Gale (2000), “Financial Contagion”, Journal of Political Economy 108(1): 1-33. 

Ari, A, S Chen and L Ratnovski (2020), “COVID-19 and non-performing loans: lesson from past crises”, VoxEU.org, 30 May.

Baldwin, R E and B Weder di Mauro (2020), Mitigating the COVID Economic Crisis: Act Fast and Do Whatever It Takes, CEPR Press.

Conesa, M, G Lotti and A Powell (2020), “Resilience and Fragility in Global Banking”, IDB Working Paper Series IDB-WP-1133.

De Haas, R and N Van Horen (2012), “International shock transmission after the Lehman Brothers collapse: Evidence from syndicated lending”, American Economic Review 102(3): 231–237.

Financial Times (2020a), “Coronavirus tracked: the latest figures as countries fight Covid-19 resurgence”, 6 October.

Financial Times (2020b), “Eurozone banks rein in lending due to pandemic worries”, 27 October.

Institute of International Finance (IIF) (2020), “LatAm Views: The QE Challenge”, 8 June.

Powell, A and L Rojas-Suarez (2020), Sound Banks for Healthy Economies, CGD-IDB Working Group Report.

Standard & Poor’s (2020), “Global Banking: Recovery Will Stretch To 2023 And Beyond”, 23 September.

The Banker (2020), “Moody’s downgrades UK banks as pandemic bites”, 23 October.

Endnotes

1 Source: https://coronavirus.jhu.edu/map.html

2 Authors’ calculations based on the IMF Fiscal Monitor Database of Country Fiscal Measures in Response to the COVID-19 Pandemic, sampled in September 2020.

1,575 Reads