Few experts, however, have said much about how to diagnose and predict--and thus possibly prevent--recurrences of the type of financial contagion that has plagued the world in the past two years. Traditional economic indicators of national health reveal little about how the fluid exchange of capital across porous national borders can infect just about every country with contagion from the outside.
Before the Asian crisis unfolded, however, a team of RAND researchers had set out to explore why some markets appear vulnerable to contagion while others do not. In Financial Crises and Contagion in Emerging Market Countries, authors Julia Lowell, C. Richard Neu, and Daochi Tong develop four "models of contagion," each with its own set of "contagion indicators" and associated prevention policies. The authors then apply their models to recent crises to demonstrate how to help fend off future disasters.
There are at least three reasons why the United States wants to predict and prevent financial contagion. First, some of the most important trade and investment partners of the United States are emerging market countries, and financial turbulence in these countries directly affects U.S. economic interests. Second, widespread failures of emerging financial markets jeopardize efforts to deregulate them and open them to foreign investors, including U.S. investors. Third, the United States wants to avoid costly bailouts of crisis-stricken nations.
The RAND authors identified eleven episodes from January 1989 to August 1997 in which stock markets crashed or currencies came under severe pressure in four or more countries nearly simultaneously. Of those eleven episodes, six appeared to be purely coincidental. Two appeared to result from a common shock, specifically the Tiananmen Square episode of June 1989 and the onset of the Gulf War in August 1990. For the remaining three episodes--the financial turmoil surrounding the U.S. stock market crash of October 1989, the Mexican peso crisis of December 1994, and the Thai baht crisis of July 1997--a statistical analysis suggested that they were indeed contagious.
Following the 1989 U.S. stock market crash, loss of investor confidence spread to world and regional financial centers as well as to some developing markets, such as those in Malaysia and Turkey. The 1994-95 Mexican peso crisis spurred the collapse of stock and currency markets in Latin America and sent fears through Asia and the financial centers. And as early as August 1997, the Thai baht crisis already had triggered widespread damage on the stock and currency markets of Indonesia, Malaysia, and the Philippines as well as in some financial centers.
Neither the statistical analyses nor traditional economic indicators, however, could explain why some countries were vulnerable to these crises while others were not. Traditional indicators for Asia, for example, had changed very little between 1994 and 1996; yet Asian countries were far more deeply affected by the fall of the Thai baht in 1997 than by the fall of the Mexican peso in 1994. According to 10 standard indicators--including gross domestic product growth, export growth, money supply growth, unemployment, inflation, real exchange rate appreciation, international reserves position, and percentage of nonperforming loans--there was no consistent pattern of economic or financial deterioration for Southeast Asian countries other than Thailand during the three years prior to the baht's collapse. In fact, many of the 1996 indicators for Indonesia, Malaysia, and the Philippines suggested economies that were still booming. Yet the financial markets of these three countries suffered severely when the baht was officially devalued in July 1997.
To help explain the apparently capricious behavior of the emerging global economy, the RAND researchers devised four models--or theories--of contagion, each with its own contagion vulnerability indicators.
In the first model, the "economic linkages" model, a financial crisis in one country spreads to other countries with which it has strong trade and investment links. The second model, "heightened awareness," suggests that investors with incomplete information may project the problems of one country onto other countries with similar national indicators, or economic "fundamentals"; this suspicion by association prompts investors to flee the suspect countries in addition to the problem country. In the "portfolio adjustment" model, portfolio managers respond to a crisis in one country by selling off assets not only in that country but also in other countries that happen to be grouped in the same investment portfolio as the problem country. Finally, the "herd behavior" model, which may be the most widely accepted view of contagion, contends that investors abandon investments largely because of what they think other investors are doing.
These four models offer different prospects for predicting and preventing contagion. Crises spread by economic linkages can most likely be predicted once the first country founders, because trade patterns are well-known and rather slow to change. However, there may be little that susceptible countries can do to blunt the impact from this kind of contagion, other than making rapid policy adjustments, such as contracting their money supplies or tightening restrictions on bank lending. Crises spread by incomplete information are more difficult to predict, but unwarranted suspicions might be prevented by better reporting and analysis of data. Portfolio adjustment crises are fairly predictable as long as managers consistently group countries according to geographic regions and as long as portfolios remain the same over time; however, preventing portfolio contagion requires the opposite--that is, portfolio diversification--which would complicate prediction. Finally, it might be impossible to predict herd behavior or "panic," even among "sophisticated" investors, and so the only defense against herd behavior might be capital controls to prevent large and undesirable capital movements (see Table 1).
The contagion indicators follow from the predictive capabilities of the models. For the "economic linkages" model, there are two indicators: strong trade and investment links with the country experiencing a crisis, and heavy trade competition with the country in crisis. For the "heightened awareness" model, there are three indicators: economic fundamentals similar to the country in crisis, potential financial or political scandals, and poor or incomplete economic data or analysis available to investors. There are two contagion indicators for the "portfolio adjustment" model: consistent membership in portfolios that also contain the country in crisis; and capital inflows that are highly leveraged, or originally borrowed from somewhere else, which makes the investors highly fearful of low returns and thus quick to pull any insecure investments. The "herd behavior" model has two possible warning indicators: dramatic capital inflows in the past, and domination of the market by less-sophisticated small retail investors and by mutual funds. Table 2 outlines these indicators and their applicability to stock market crashes and currency devaluations.
To determine the predictive power of the models, the authors tested them against three recent, real-world financial crises:
In contrast, South Africa in 1996 was a true "loner" country, both as a result of its geographical isolation from other developed markets and because of its former status as an international political pariah. For those reasons, according to the contagion models, the currency crisis in South Africa remained isolated. With no strong economic or financial ties to other countries, with a unique recent economic and political history, with no involvement in regional investment portfolios, and with limited attraction for retail investors, South Africa's crisis passed almost unnoticed by investors and other emerging markets.
Most worrisome of all, the contagion of Southeast Asia has continued to spread across oceans and markets. This case provides perhaps the clearest example of the failure of traditional indicators, because all four contagion indicators were relevant at the outset: strong economic linkages within Southeast Asia; similar economic fundamentals, including poor export performance and a large share of nonperforming loans in banking and real estate; common membership in regional investment portfolios; and the potential for being deserted just as quickly as they had been embraced by fickle investors.
In Southeast Asia, industries such as semiconductors and textiles are highly concentrated, and firms there tend to see each other as prime competitors. Investors may well have worried that firms based in Indonesia, Malaysia, and the Philippines could no longer compete against Thai firms after the devaluation of the baht. Financial connections within the region are also strong: For example, Singapore banks with extensive loans in Thailand and Malaysia immediately felt the impact of the collapse. Other Southeast Asian countries also shared Thailand's problems in real estate and banking, which could have prodded investors to worry that additional bank scandals lurked in nearby closets. At the same time, Southeast Asian countries are heavily grouped in regional investment portfolios. Finally, Southeast Asia was the darling of international investors throughout the early 1990s, many of whom were small-scale retail investors. Such sudden, huge, short-term capital inflows are highly vulnerable to changes in investor sentiment--changes probably triggered by events in Thailand.
Externally generated crises very well might be beyond the power of individual governments to prevent, especially when the crises involve multiple kinds of contagion, as in Southeast Asia. Nonetheless, the RAND researchers contend, it is possible to predict the ways that contagions will spread, given better indicators. And if better warning signals can be developed, then national leaders might, at the very least, have a better idea whether their markets are vulnerable to contagion and what might be done to inoculate their economies.
The APEC Model of International Economic Cooperation: Assessing Its Value to the United States, Julia Lowell, RAND/ P-8028, 1998, 23 pp., $5.00.
U.S. Economic Unilateralism: Implications for Pacific Rim Countries, Julia Lowell, RAND/P-8029, 1998, 27 pp., $5.00.