Contagion

Contagion in a broad sense has been studied as the propagation of an initial adverse macroeconomic shock from one market or economy to another. It has been characterized by robust comovements or excess positive cross-country correlations in macro-financial indicators (for instance, interest rates and sovereign spreads), beyond what can be explained by fundamental economic variables (see Bekaert et al., 2005). Most empirical literature still rests on the notion of "shift contagion"; that is, significant variations in pre-existing cross-market linkages (for example, correlations and speculative attacks) or changes in the transmission mechanism between two markets or economies in crisis periods (see Forbes and Rigobon, 2002).

The measurement of contagion is best echoed in the international portfolio theory (IPT). According to the IPT, taking a Chinese investor as an example, international port- folio investments in advanced foreign markets (therefore dissimilar or less integrated) like the United States are highly desirable, as these drive inter-country correlations between bonds and stocks even further down, thereby optimizing risk reduction and maximizing asset returns (Solnik and McLeavey, 2004). The intuition is that most adverse macro- economic shocks are country-specific, such that financial markets in different economies display low correlations. The presence of contagion, therefore, apprehends this reason- ing. Here, it is worth noting that Chinese and euro-area portfolio flows into US markets have been deemed a factor of the global crisis that erupted in 2008 (see Bernanke's "global savings glut" hypothesis). In another sense, investors can hedge or diversify home-country risks through direct or capital investments in high-growth economies (for instance, US lending and portfolio flows to Mexico in the early 1990s that preceded the peso devaluation of 1994 and the subsequent "tequila" crisis).

One misconception of international diversification is entailed in the correlation breakdown theory (CBT). The CBT derives that, through higher market interaction (interdependence and/or contagion), correlations tend to "break down" exactly when they are needed most, such that the benefits of international diversification cannot manifest in crisis periods. Following this thought, the leading attempt to capture con- tagion has been to identify the "correlation breakdown"; that is, when the correlations of tranquil periods become unprecedentedly higher in periods of turmoil (as in crisis- contingent models). This is simply done by isolating the autocorrelation coefficients between several international markets (for instance, equity and bond markets) during major global spillovers (see King and Wadhwani, 1990; Calvo and Reinhart, 1996; Baig and Goldfajn, 1998).

Some authors have conceptualized contagion as the spillover of volatility between markets or economies and have adopted generalized autoregressive conditional het- eroscedacity (GARCH) models in their empirical tests, while a third literature stream testing for cointegrating relationships employed longer time periods. Probit models later emerged to examine the contagious effects of exogenous events to an economy (see Forbes and Rigobon, 2002, for a review of empirical schools). The idea with factor models is to detect the autocorrelations in model residuals in line with the extreme-value theory (see Bekaert et al., 2012). Phylaktis and Xia (2009) add that some industries are more resilient than others, in the sense that they continue to emulate the merits of international diversification during contagious crises.

The first problem with the measurement of contagion is that almost all definitions and empirical studies have been prudent in specifying the channels through which con- tagion occurs. Further, if contagion measures a significant shift in pre-existing cross- market linkages, why do markets with very few linkages exhibit such a very high degree of comovement in the first place? South Korea, for example, was hit by the 1997 Asian crisis despite having very limited links with other Asian economies. Generally, theories to explain the channels in the transmission of shocks range from multiple equilibria, endogenous liquidity to other phenomena like the "wake-up call hypothesis".

A second class of ambiguities is driven by issues with methodological aspects, and is exacerbated by the lack of a unanimous conceptual framework. The first problem in this regard is the endogeneity of asset prices. The second dispute is on how to properly account for heteroscedacity: the correlation function used in empirical models is always increasing in the variance of the underlying asset return, implying that volatilities are naturally higher during crisis periods, making it difficult to attribute any robust comove- ments to contagion. After correcting for heteroscedacity, Forbes and Rigobon (2002) found no evidence of contagion in the 1994 Mexican peso crisis, the 1987 US stock market crash, nor the 1997 East Asian crisis. They argued that higher correlations of crisis periods are simply a continuation of pre-existing real linkages and are not suffi- ciently significant to represent contagious effects.

Another complication is the omission of relevant variables: if any macroeconomic indi- cators that would exhibit strong autocorrelation are precluded from models, the statisti- cal results become significantly biased. The application of different event windows and the subjective definition of idiosyncratic shocks create further divergence with results. A high level of econometric creativity can be observed in this regard. One reason is to cir- cumvent problems with the limited frequency of the availability of macroeconomic data. Forbes and Rigobon (2002) criticized the relevance of previous studies, as authors have resort to historical periods, in order to avoid problems associated with public safety nets like the "lender of last resort".

Finally, Beirne and Gieck (2012) note that the transmission of shocks depends on their origin. For instance, while US equity market shocks are identically transmitted to regional equity markets, a negative euro-area stock market shock would transcribe into favourable outcomes in regional markets. It is important to note that contagion, if considered to be independent of economic fundamentals, might become an obstacle to optimal policy. That is, if several crises cannot be explained by trade links, some economists and many financial institutions in trouble would strongly argue that a no-bail-out strategy by policy institutions would allow for spillovers to other economic systems even if the latter were fundamentally delinked from the one in trouble.

See also:

Bank run; Financial crisis; Lender of last resort.