The feedback loop between fiscal and financial instability has been at the core of the recent turmoil in Europe (Acharya et al. 2014). In some countries, systemic banking crises triggered fiscal distress due to the magnitude of bank rescue operations (for example, in Ireland). In others, substantial sovereign debt tensions, leading to successive sovereign downgrades, severely weakened domestic financial systems (for example, in Greece). Against this background, and despite the fact that the combination of fiscal and financial distress is nothing new to emerging markets, it is surprising that the large literature looking at how different types of crises combine (the so-called ‘twin crises’ literature) has only recently begun to examine the links between banking and sovereign debt crises. Concerning emerging markets, two notable exceptions are Panizza and Borenzstein (2008) and Reinhart and Rogoff (2011). These papers document the existence of feedback between banking and sovereign crises, but do not study the channels of transmission formally.1 In Balteanu and Erce (2014), we fill this gap by studying the behaviour of a set of macro-financial variables through which feedback loops of fiscal and financial stress may materialise. New to the literature, we differentiate twin events according to whether they are triggered by a banking or a debt crisis.
Definitions, data, and methodology
The study covers 78 emerging countries from 1975 to 2007. Sovereign crises are defined using Standard and Poor’s data. The banking crises are the ‘systemic events’ catalogued in Laeven and Valencia (2013). Combining both datasets, we isolate four types of events: ‘single’ banking crises; ‘single’ sovereign debt crises; ‘twin bank-debt’ crises, where a banking crisis is followed by a sovereign one within three years; and ‘twin debt-bank’ crises, where a sovereign crisis is followed by a banking one. Using these definitions, there are 140 distinct crises in the sample, of which 51 are single banking crises, 61 are single debt crises, 15 are twin bank-debt crises, and are 13 twin debt-bank crises. Our choice of macro-financial variables is guided by the channels of transmission we aim to assess. These include direct balance sheet effects (banks’ public debt holdings, central bank credit, and public recapitalisations), as well as other indirect ways through which these crises, and the policies deployed to address them, impact the economy.2
Following Gourinchas and Obstfeld (2011), we use an event analysis methodology to estimate how the conditional expectation of each variable depends on the temporal distance from each type of crisis, given the proximity of other crises, and relative to a ‘tranquil times’ (non-crisis) baseline, within a six-year window around the events. The fact that the ‘tranquil times’ baseline is common to all events makes the comparison among coefficients straightforward and allows us to compare the dynamics of each variable around crises of different types. Figures 1–8 present a selection of the results in Balteanu and Erce (2014), where we plot the estimated coefficients obtained for each variable throughout the crisis window.
What comes with banking crises that trigger sovereign default?
The interplay between banks’ and both central bank’s and government’s balance sheets is systematically different in banking (B) and twin bank-debt crises (BD). Ahead of single events, low exposures to government combine with high liquidity support from the monetary authority. In the aftermath, liquidity support drops substantially and banks’ holdings of public debt start to accumulate. Ahead of BD events, the fast and substantial accumulation of government paper on banks’ balance sheets combines with slow-growing central bank support, while in the aftermath, the increase in public claims moderates and liquidity support shoots up. These diverging patterns could be due to differences in the initial shock, policies implemented by the central bank and government, structural features of the banking sector, or a combination of all these factors. In fact, around BD events, banking systems are significantly larger (needing more support in situations of stress), the impact of the crisis on growth is more severe (also providing incentives to the authorities to intervene) and the loss in foreign investors’ confidence is more significant. In ‘single’ episodes, banks start to size down already ahead of the crisis, and continue to do so as the crisis unfolds. In contrast, in ‘twin’ episodes, asset downsizing starts late in the banking crisis, and the process is more gradual. This might indicate that the policy response is to try to keep the banking sector afloat, postponing deleveraging until the crisis engulfs the public sector as well. Table 1 shows that in BD events the authorities fund much larger rescue packages than in any other type of banking crises.
Table 1. Intensity of banking crises: static indicators (as in Laeven and Valencia 2013)
Sources: Laeven and Valencia (2013), S&P, and authors’ calculations.
Notes: ‘NPL’ refers to non-performing loans. Change in number of banks refers to the change on the first three years of the crisis. Fiscal and recapitalisation costs are measured as % of GDP.
While the two events occur against similar initial debt and budget positions, public finances diverge once banking crises are underway. Banking crises that are part of ‘twin’ events are associated with a sharper increase in the budget deficit and a larger accumulation of public debt, which suggests that they put more strains on government finances than ‘single’ banking events.
What happens when a sovereign debt crisis ends in a banking crisis?
Regarding the differences between ‘single’ debt (D) and ‘twin debt-bank’ (DB) crises, we find that, ahead of the latter, banks are more exposed to the government, and the pace of increase in their public debt holdings is faster. The amount of liquidity support provided by the central bank is significantly larger than ‘normal’ around the two types of episodes. Nevertheless, while liquidity support is flat throughout D crises, it increases dramatically during DB ones. In addition, the banking sector is smaller in DB events than in D ones. These differences could indicate that ‘twin’ defaults are more damaging to banks’ balance sheets, and also may leave the sovereign with less margin to support the banking sector. Finally, around DB events there is little foreign capital flowing into the economy.
While the state of public finances is similar ahead of D and DB events, public expenditure is cut more drastically (and public debt drops faster) in the aftermath of ‘twin’ defaults. This could point to either a lack of fiscal space, or the adoption of a more austere stabilisation package, which may negatively affect the economy in the short run. In fact, DB defaults have a larger negative impact on growth, and the recovery is slower. These growth dynamics are accompanied by inflation rates that fall more markedly during ‘twin’ events than during ‘single’ ones – a further indication of tight austerity being implemented in the aftermath of DB defaults.
Apart from the systematic differences between ‘single’ and ‘twin’ crises, our analysis also shows that ‘twin’ crises themselves are far from being homogenous events. By taking into account the different sequence of crises during ‘twin’ episodes, we are able to uncover significant differences between the two types of ‘twin’ crises, both in terms of levels and dynamics, which would have otherwise gone unnoticed – such as those of budget deficits, inflation rates, credit to the private sector, or capital flows. This is a relevant result, given that the twin-crises literature does not typically distinguish twin crises according to the original shock.
By providing a detailed understanding of the economic dynamics around different crisis episodes, the event study discussed in this column can inform the design of theoretical models. In fact, a number of empirical facts usually associated with either ‘banking’ or ‘debt’ crises in the literature are to be found in ‘twin’ events only. Moreover, these results show that considering the sequence of crises within ‘twin’ events is important for understanding their transmission channels and economic consequences.
Disclaimer: The views in this column are the authors’ and should not be reported as those of the European Stability Mechanism, the Bank of Spain, or the Euro-System.
Acharya, V, I Drechsler, and P Schnabl (2014), “A Pyrrhic Victory: Bank Bailouts and Sovereign Credit Risk”, Journal of Finance, forthcoming.
Balteanu, I and A Erce (2014), “Banking Crises and Sovereign Defaults in Emerging Markets: Exploring the Links”, Bank of Spain Working Paper 1414.
Laeven, L and F Valencia (2013), “Systemic Banking Crises Database”, IMF Economic Review 61(2): 225–270.
Obstfeld, M and P O Gourinchas (2012), “Stories of the Twentieth Century for the Twenty-First”, American Economic Journal: Macroeconomics 4(1): 226–265.
Panizza, U and E Borensztein (2008), “The Costs of Sovereign Default”, IMF Working Paper 08/238.
Reinhart, C M and K S Rogoff (2011), “From Financial Crash to Debt Crisis”, American Economic Review 101(5): 1676–1706.
Annex: Macroeconomic dynamics around “single” and “twin” events
Figure 1. Credit from the central bank (% assets)
Figure 2. Claims on government (% GDP)
Figure 3. Public debt (%GDP)
Figure 4. Portfolio capital inflows (% GDP)
Figure 5. Credit from the central bank (% assets)
Figure 6. Claims on government (% GDP)
Figure 7. Budget expense (%GDP)
Figure 8. GDP growth (%)
1 Reinhart and Rogoff (2011) document that banking crises significantly predict sovereign defaults, but find no evidence of the opposite link. Conversely, Panizza and Borenzstein (2008) find that sovereign default can trigger banking crises, but do not find the opposite.
2 We consider banks’ characteristics (assets, deposits, and private sector credit), public finances (budget balance, revenues, expenses, and debt); and other macroeconomic factors (growth, inflation, capital flows, and external debt).