What dynamics and level of integration would make for the most efficient and stable global capital market? It’s an ambitious question, but as a start let’s examine the patterns around the financial crisis. The increase in international capital flows before 2008 and the decline afterwards have been widely documented. Changes in cross-border banking flows made up a substantial part of this rise and fall.

Breaking down the data by country reveals two main trends. Banks in developed countries, particularly in Europe and the United States, retrenched and cross-border claims fell, but banks based in emerging economies actually increased cross-border lending. Another interesting contrast is that while global integration in banking declined, regional integration increased. 

The crisis particularly affected cross-border flows in the eurozone, as research from the Bank for International Settlements (2012) and Bologna and Caccavio (2014) found. Gross consolidated foreign claims of eurozone countries decreased by 35 percent from the peak in March 2008 to the trough in June 2012. Claims from eurozone countries on other eurozone countries fell by 40 percent compared to 32 percent for claims from eurozone countries to non-eurozone countries. These average numbers hide a huge variation; the changes in individual countries range from a fall of 8 percent for Portuguese banks to 80 percent for Belgian banks.

After the crisis the absolute number of banks operating in foreign countries fell, but the foreign banks share remained the same, at about 35 percent. Although the total number of foreign banks entering new countries fell to about one fifth of what it was pre-crisis, the number of domestic banks also fell. Claessens and VanHoren (2014) document the aggregate data and the sharp differences across countries. Again, there is significant variation. In 66 host countries the number of foreign banks fell, but in 48 it increased. Banks from emerging countries were responsible for over two thirds of new entries. In addition, the average intra-regional share increased by 5 percent.

Emerging markets banks were expanding abroad even before the crisis, and this trend has continued. Their role increased as well-capitalized emerging-country banks were able to purchase interests from developed-country banks in the aftermath of the crisis. Claessens and Van Horen give the examples of Russia’s Sberbank buying the Central and Eastern European subsidiaries of Austria’s Volksbank; Chile’s Corpbanca buying Santander’s Colombian; and HSBC selling its interests in Costa Rica, El Salvador, and Honduras to Banco Davivienda of Colombia. Overall, emerging-country banks increased their share of foreign banks assets from 4 percent to 8 percent, while the banks within the Organisation for Economic Co-operation and Development (OECD) decreased their share by 6 percent.

Academic literature has emphasized four explanations for the changes in cross-border lending, which we will now consider in turn.

The Crisis Itself Changed the Nature of Cross-Border Banking
Considerable evidence demonstrates the crisis itself significantly affected the willingness of banks to lend across borders. Claessens and Van Horen (2014) found that banks with headquarters located in a crisis country are more likely to have exited from any host country. De Haas and Lelyveld (2010) found that foreign banks were more willing to support their subsidiaries when the host country was in crisis and the bank’s home country was not. Research from BIS (2012) and Bologna and Caccavio (2014) suggests that the eurozone has been particularly affected due to the difficulty of hedging the risk of a eurozone break-up, except by balancing assets and liabilities country by country.

Claessens and Van Horen (2014) found that banks are more likely to exit: when the host country is less developed; when the home country is more developed; when home country banks have a small share of the host country market; and when the distance between the two countries is large. Not surprisingly, entry is higher when real GDP growth is greater—and this turns out to be the main determining factor. Growth in assets of banks that are already established in a foreign country is driven by a wide range of host country characteristics and conditions.

Changes in Regulation Affected the Willingness of Banks to Lend Across Borders
In theory the impact of new regulation on cross-border lending can go either way, as Houston, Lin and Ma (2012) point out. On one hand, banks may be attracted to cross-border lending if the host country has less restrictive regulations than the home country. On the other, stricter regulatory requirements in the home country may make foreign lending more expensive. Bremus and Fratzscher (2014) find that increasing capital requirements in the home country reduces cross-border claims. They argue that regulators must implement new rules across countries in a transparent and harmonized way in order to avoid further reductions in cross-border banking.

Central Bank Actions Affected International Banking
Many have argued that loose monetary policy since the crisis has been an important factor in cross-border flows. Bruno and Shin (2014) develop a model of global liquidity built around the operation of global banks. These finance cross-border lending to local banks by borrowing U.S. dollars from money market funds in financial centers. The local banks lend to the ultimate borrowers in U.S. dollars. The local banks then invest in local currency assets and so create a currency mismatch. When the local currency appreciates (that is, the U.S. dollar depreciates) the ultimate borrowers’ balance sheets become stronger, and local banks can lend more to them. Local currency appreciation thus leads to increased risk-taking by banks. When there is local currency depreciation, on the other hand, the process reverses. Banks deleverage and cut loans, threatening financial stability.

Loosening of U.S. monetary policy dollar can lead to depreciation against some currencies and increased cross-border lending. Appreciation of currencies can lead to reductions in cross-border lending. Bruno and Shin use data from 46 countries and find support for their model. This theory is one potential explanation for the wide heterogeneity, since the crisis, in cross-border bank lending to emerging countries, with lending to some countries going up and to many other countries going down. In many emerging countries, local currencies have strengthened against the U.S. dollar, while in many countries currencies have weakened.

The Resolution of Cross-Border Banks Proved Problematic
The lack of a global resolution mechanism has been a significant determinant of cross-border lending, according to much research, e.g. Cerutti, Claessens and McGuire (2012)—again, particularly so in the eurozone, as Manna (2011) points out. It is a commonplace to say that banks are global in life, but national in death. As a result, national regulators have an incentive to limit cross-border lending (The Economist, 2012). After the default of Lehman Brothers, this factor seems to have become significantly more important.

What Should We Look at Next?
This overview suggests some broad trends, but the data on cross-border banking is complex and interpreting it is challenging. Drawing firm conclusions is premature at best and foolhardy at worst. The currently available data is just a rough measure of cross-border banking activity. The size of the cross-border flows is important, but the data hide a great deal of qualitative information about the many activities that banks engage in. Apart from correlations that reveal the ebb and flow of lending from countries of different sizes, locations, and levels of development, there is little hard data that bears directly on the causal factors driving these changes. To conclude, we outline some of the most interesting and important questions posed by the data and the kinds of research that could settle the related policy questions.

Are capital flows the right measure of globalization, or are there qualitative dimensions that should be taken into account?
Efficient risk-sharing implies a particular pattern of capital flows among countries, where countries in recession receive flows from countries in the boom. Macroeconomists have documented the failure of capital markets to provide this kind of insurance, but the possibility reminds us that it matters, for the allocation of risk and resources, whether Country A is receiving funds from Country B or Country C.

Does greater regional integration lead to greater stability?
Looking at the big picture of global financial stability, the efficiency of cross-border lending can only be evaluated if one knows the risks being shared, the correlation of those risks, the productivity of firms and industries being funded, the cost and maturity of the debt, the types of controls exercised by the lender through covenants and monitoring, and other factors. Because this data has not been collected, the jury is still out.

Does the retrenchment of the banks in developed countries make the global system more stable?
The retrenchment by banks based in developed countries and the expansion of banks based in emerging countries represent a replacement of one set of banks by another, but it may also represent a shift from one set of borrowers to another. The aggregate data in the studies to date does not reveal what kind of reallocation has occurred. Whether it has increased the efficiency of the global financial system is anyone’s guess.

Even if improved regulation and higher standards in developed countries makes individual banks stronger, does greater fragmentation of the global system add to stability or simply reduce risk-sharing?
As recent theoretical research on financial networks makes clear, greater connectivity can promote diversification, if the shocks are not too great, and this is one of the virtues of globalization when it works well. But greater connectivity can be a transmitter of shocks that, if the shocks are large enough, will destabilize the system. Similarly, greater integration of regional economies may concentrate risk and leave countries more exposed to shocks emanating from their neighbors, whose economies are highly correlated with their own.

Another qualitative dimension of the aggregate changes observed in the data relates to the services provided by different banks. Financial institutions from developed countries, especially the very largest banks, provide a range of services that cannot be duplicated by most banks in emerging markets. It may be that these services are not needed in some parts of the world are best restricted to their host countries. On the other hand, large banks like to say that their customers want to deal with a bank that can provide “one-stop shopping” for financial services and can provide these services globally, wherever their customer, often a multi-national firm, operates. A decrease in globalization could involve a cost born by those corporate customers.

What explains the changes in cross-border lending that we have observed?
We have mentioned a number of theories. Each of them may be relevant in some countries and not in others. One calls for stricter macro-prudential regulation in developed countries. As banks have been forced to raise more capital and hold more high quality liquid assets, or else shrink their balance sheets and obtain more long term funding, they have found it expedient to reduce their involvement in emerging markets. The Great Recession reduced growth and profits in many parts of the world and that has had knock-on effect on borrowing. The sovereign debt crisis in the eurozone has had a direct impact on cross-border lending as banks in the southern periphery have loaded up on the debt of their own government, and banks in the northern core have shed the debt of the southern. But the stress that was felt by the eurozone banking system must have had wider repercussions on lending to emerging markets or indeed to any country outside the eurozone.

Conclusion
The issue of causation is interesting and important in itself. This also depends on the qualitative dimension of the changes in global capital flows, because banks in different parts of the world are not perfect substitutes and because lending from different parts of the world has different implications for risk sharing and stability. Much more research is needed before we can understand the implications, for efficiency and stability, of the changing tides and eddies of the global capital market.