The collapse of global equity markets between August 2007 and March 2009 has been part of the most severe global crisis since the Great Depression. While the crisis initially had its origin in the US in a relatively small market segment, the subprime mortgage market, it rapidly spread across virtually all economies, with many countries experiencing even sharper equity market crashes than the US. This makes this episode an ideal laboratory to revisit the debate about the presence and sources of contagion in global equity markets.
The definition of the term contagion remains highly controversial (see eg Forbes and Rigobon 2002). A relatively agnostic proposition derives from the asset pricing literature and defines contagion as co-movements of equity returns above and beyond what can be explained by fundamentals. Using this approach, we analyse the equity market transmission of the 2007-09 global financial crisis to sector portfolios in 50 countries through three channels: via US equity markets, via the global financial sector and via domestic equity markets (see Bekaert et al 2011). This approach thus identifies three different types of contagion, linked to each of these channels.
A first striking feature emerging from the analysis is that—perhaps contrary to conventional wisdom—contagion from the US or the global financial sector played little role during the crisis, while contagion mainly occurred through domestic channels. This implies that investors, while discriminating how news should affect equity prices across countries, were discriminating much less across stocks within countries.
The lower panel of Table 1 reports estimates of the interdependence of equity returns during non-crisis times, by displaying the average of the elasticities (“betas”) of equity returns to the three factors. All three elasticities are similar, at about 0.4-0.5, implying that a 1% increase in a given factor return is associated with an average 0.4%-0.5% increase in country-sector-level returns.
Yet things turned out to be very different during the 2007-09 global financial crisis. The upper panel of Table 1 reports the average of the contagion elasticities (“gammas”) to the three factors during the crisis. Most strikingly, contagion during the 2007-09 crisis seems to have been primarily domestic in nature. The estimate for the domestic contagion elasticity, at around 0.25 on average, is indeed much larger than the analogous estimates for the US (at around 0.13) and the global financial sector (at around 0.06).
This points to evidence that contagion was mostly domestic in nature and did not stem systematically from the US or the global financial sector. Interestingly, this feature seems quite specific to the 2007-09 crisis, as there is no evidence that domestic contagion played a role in past crises, such as the 1998 LTCM crisis or the 2000-02 bust of the TMT bubble.
Some may object that the proposed approach will not necessarily be able to explain much of the huge equity collapse that most firms and countries experienced during the crisis. However, Figure 1 demonstrates the good fit of the model by showing the close relationship between actual total returns (aggregated to the country level) over the crisis period and the predicted values from the empirical model. The reason for this good fit is in particular the model’s flexibility, since it allows elasticities to differ across country-sector portfolios.
The second main finding emerges from the analysis of the main determinants of contagion—which turn out to be mainly the strength of countries’ fundamentals and the quality of their institutions. Specifically, countries with poor macroeconomic fundamentals, high sovereign risk, and poor institutions experienced by far the largest equity market declines and were most severely affected by contagion. The fact that domestic fundamentals played such a key role in the transmission of the crisis is reminiscent of the old “wake-up call” hypothesis, whereby a crisis initially restricted to one market segment or country provides new information that prompts investors to reassess the vulnerability of other market segments or countries, ultimately spreading the crisis across markets and borders (see Goldstein 1998; Goldstein et al 2000).1
By contrast, the external exposure via trade or financial linkages explains little of the differences in severity of contagion affecting firms and countries during the crisis. This is at odds with the view that globalisation or worldwide economic and financial interdependence was a key channel in the transmission of the crisis, in that those economies that are highly integrated globally, through trade and financial linkages, were indeed not necessarily those that were hit hardest by the crisis.
In summary, the financial crisis did not seem to have spread indiscriminately across countries and economic sectors. In particular portfolios in countries with weak economic fundamentals experienced more contagion and were therefore overall more severely affected by the global financial crisis. The irony of this perhaps most-global crisis ever is that differences in external exposure played after all such a small role in determining relative equity market performance and contagion. Instead, investors refocused primarily on country-specific characteristics and punished markets with poor macroeconomic fundamentals, policies, and institutions.
Overall, this evidence supports the recent efforts by policymakers and international organisations to better understand macroprudential risks and perhaps institute a closer surveillance of such risks both at a country level and at the global level.
Table 1. Equity market contagion and interdependence estimates
Note: The table reports average estimates of the degree of contagion (“gamma”) and interdependence (“beta”) across 415 country-sector portfolios equity returns relative to three factors: the global financial sector (G), US equity markets (U) and domestic equity markets (D). The model is estimated with weekly data over January 1995-March 2009 (crisis period: August 2007-March 2009), allowing errors to be clustered by country. ***, **, and * indicate statistical significance at the 1%, 5%, and 10% levels, respectively.
Source: Bekaert, Ehrmann, Fratzscher, and Mehl (2011).
Figure 1. Goodness of fit—Contagion model
Note: The figure shows the cumulated actual equity market returns for 415 country-sector portfolios, aggregated at the country level, over the entire crisis period (August 2007—March 2009) against the fitted cumulated returns from the contagion model The dashed line shows the 45 degree line, the solid line represents a regression line.
Source: Bekaert, Ehrmann, Fratzscher and Mehl (2011).
Bekaert, G, M Ehrmann, M Fratzscher, and A Mehl (2011), “Global crises and equity market contagion”, NBER Working Paper, No. 17121.
Forbes, K and R Rigobon (2002), “No contagion, only interdependence: measuring stock market comovements”, Journal of Finance, 57(5):2223-2261.
Goldstein, M (1998), “The Asian financial crisis causes, cures, and systematic implications”, Institute for International Economics, June.
Goldstein, M, GL Kaminsky, and CM Reinhart (2000), “Assessing financial vulnerability: Developing an early warning system for emerging markets”, Institute for International Economies.
1 This term was coined by Goldstein (1998) in the wake of the Asian financial crisis, with the Thai currency crisis of 1997 acting as a “wake-up call” for international investors who eventually recognised that the so-called “Asian miracle” of the time was rather an “Asian mirage”, which ultimately led to a reassessment of the creditworthiness of Hong Kong, Indonesia, Korea, Malaysia, and Singapore.