The contributions to systemic stress of financial interactions between the US and Europe

Abstract

Understanding the connectivity of international financial markets is critical to understanding the origination and propagation of financial crises. This study investigates the contribution of US and European exchange rate interactions to overall stress in the US financial system from 1992 to 2013. The impacts of these interactions are assessed using a financial stress index that aggregates measures of national and international stresses. There are three main findings for the sample period. First, we find that European influences on US financial stress have increased. Second, observing several structural breaks with changing correlation and Granger causality patterns, we find that the euro and the British pound have contributed varying levels of stress. Third, we find that stress in US markets tends to spill over into European markets, while the reverse influences are of lesser importance. These findings have important implications for supervisors in international markets. Understanding the amplifying or attenuating feedback effects from international connectivity provides valuable insight into the development of macroprudential policies.

Keywords:

• financial stress
• financial stress index
• supervision
• systemic risk
• risk contributions

JEL Classification:

• G01
• G10
• C22
• C52
• E32
• E44
• E6

Acknowledgements

The authors would like to thank the participants at the 2014 Portuguese Finance Network conference (Vilamoura, June 2014) and the 2015 INFINITY conference on International Finance (Ljubljana, June 2015) for their supportive comments. They specially express their gratitude to Amanda Janosko for her valuable research support. The authors also thank the reviewers and the journal editors for constructive suggestions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. The beginning of the crises considered in this study is generally associated with the collapse of Lehman Brothers in September 2008. While the crisis first originated in the US mortgage market, it was further affected by the European markets and has merged with the effects of the European debt crisis.

2. The suitability of spreads as an expression of stress on financial markets (default risk and liquidity risk) is discussed from Illing and Liu (2006, 251–253) and Cheung, Fung, and Tsai (2010, 85–86).

3. It is also necessary to mention that the concepts of systemic financial stress and systemic financial risk – although often applied similarly – are not identical. Systemic stress is manifested through financial imbalances that threaten the system, whereas systemic risk is estimated as the joint probability that firms will default as a result of stress.

4. This FSI refers to the Cleveland Financial Stress Index (CFSI), developed by the authors at the Federal Reserve Bank of Cleveland (Oet et al. 2011; Oet, Dooley, and Ong 2015). The CFSI is well established and recognized in the literature (Oet et al. 2013; e.g. applied from Chau and Deesomsak 2014; Corbet 2014).

5. The basic description of variables and calculations follows Oet et al. (2011) and Oet, Dooley, and Ong (2015).

6. The currency crashes and covered interest spreads as sources of international stress are parsed for seven of the G20 economies (Oet, Dooley, and Ong 2015). These countries are Australia (AUS), Canada CAN), the eurozone (EUZ), Great Britain (GBR), Japan (JAP), Mexico (MEX), and South Africa (SAF). The countries have been chosen because their exchange rates are floating and because they have the largest trade balances with the US. Data on the trade between the US and these countries can be retrieved from the United States Census Bureau ‘U.S. International Trade in Goods and Services Report.'

7. The Z.1 ‘Financial Accounts of the United States' report comprises the Flow of Funds, Balance Sheets, and Integrated Macroeconomic Accounts.

8. The FFIEC E.16 report is the ‘Country Exposure Lending Survey and Country Exposure Information Report’.

9. VSTO has been introduced in 2005 and data goes back to 01/04/1999; the VFTS has been launched in 2008 with data going back to 01/04/2000.

10. With regard to their use in practice, components based on daily data would be preferable. Here, weekly data have been chosen for a more transparent representation. However, daily data is available and calculations have also been performed on a daily basis. The results for descriptive statistics, correlations, and break points both for the FXM components and the overall FSI are mostly identical for weekly (1148 observations) and daily data (8036 observations).

11. The rates within the EMS were fixed in 1999, and the currency was introduced as cash in 2002.

12. Since this study refers to aggregate stress in US markets, the breakpoints, according to the FSI series, are chosen as reference dates.

13. As Lee, Lin, and Wu (2002) and Maziarz (2015) point out, Granger analysis can reject the null hypothesis of non-causality in one direction, in two directions or not reject it. In these cases, different interpretations and therefore different causalities are possible.