How to measure financial integration?

Alexandra Zeitz (Global Economic Governance Programme, University of Oxford)

Speaker: Professor Jennifer Corbett, Australian National University

Chair: Robin McConnachie, Oxford Analytica

Though we know financial integration is deepening, we still know woefully little about how to measure or observe financial integration. In a seminar at PEFM in June, Professor Jennifer Corbett introduced a new research project that uses network analysis to try to address this gap and reveal patterns of financial interconnection.

The motivation for Corbett’s research is clear, given her focus on the East Asian economy and financial system. In the aftermath of the 1997 Asian financial crisis, many regional analysts concluded that the risks had stemmed from dependence on foreign intermediation. The policy prescription was thus to increase regional financial integration, especially in local currencies, with deeper, more tightly integrated markets seen as a driver of stability.

But, as Corbett pointed out in her presentation, the repercussions of tighter integration remain understudied. In part, this is because existing models of financial integration remain simplistic.

 Theories of financial integration have largely borrowed from models of trade integration. Frequently used measures include capital flow data that mimic trade openness indicators. Yet Corbett argues that these indicators are unlikely to capture potential causes of instability, since they’re unable to identify channels for contagion or those markets that are key nodes for risk.

The ultimate aim of Corbett’s research is to stoke a policy debate about the right architecture needed to manage the growing complexity of interconnections in the Asian region’s financial sectors. To do that, she and her co-authors are turning to network models for more complex and nuanced pictures of integration.

Network models are becoming increasingly popular in the social sciences, used to analyze complex interactions between many actors. For a short introduction, see this piece. Network models can reveal the pattern of interactions across a system, showing whether interconnections are random or if certain nodes are central to the system.

Using Bank for International Settlements (BIS) data on cross-border flows, Corbett and her co-authors have initial results on the global and regional interconnections. Their financial interconnectedness index, which takes into account the intensity of each country’s links to other countries in the system, reveals the nodal importance of particular countries.

Unsurprisingly, the US tops the list in the global network, followed by the UK, Germany and France. But, the Netherlands (5), Spain (9) and Luxembourg (10) also make the top ten list (places 6-8 are taken by Japan, China and Hong Kong).

The diagram of the global network of financial interconnection shows it to be highly skewed, with a small number of hugely connected countries and a large periphery that is mostly disconnected from the network. These types of lopsided networks, called ‘scale-free’, are known to particularly vulnerable to targeted shocks. A random shock may do little damage since so many countries are disconnected from the system, but a shock in one of the central nodes has many paths to spread into the system as a whole.

Looking at just the Asian regional network uncovers telling differences from the global network of financial interconnection. Among the top five interconnected, there aren’t the large gaps that exist between the US, UK and all other countries in the global network. Instead, Hong Kong, Singapore, China, Australia and Japan are all at similar levels of interconnection.

Strikingly, China has been becoming more and more closely interconnected since 2009. This result surprised Corbett, and also other participants in the seminar, indicating the country has become central to the region’s financial system in spite of enduring capital account restrictions.

The results are preliminary, and Corbett stressed that they may change in further analysis. There are data concerns, as participants in the seminar pointed out, since reporting to the BIS is patchy. The cross-border flow data is also exclusively bank data, but, as several audience members commented, integration and vulnerability may be more likely to be driven by non-bank flows.

Corbett took on board the audience suggestion that including the US in the Asian regional model may reveal how closely connected Asian financial sectors still are to the States, almost twenty years on from the Asian financial crisis.

The analysis may be in its early stages, but the discussion at the seminar showed the clear potential for policy impact. By providing a more plausible measure of interconnection, and showing the density of regional financial integration, this research can demonstrate that financial stability cannot rely on individual countries’ regulatory and supervisory efforts alone. Coordination is sorely needed: home-host cooperation, cross-border resolution agreements, regional colleges of supervisors.  One can hope that that this new research will help move the policy discussion in the right direction.