A key premise of this paper is that corporate crisis dynamics are associated with contractions of GDP more severe than the usual cyclical downturns. The real GDP changes for the nine episodes examined in this paper, measured as the sum of negative consecutive real annual GDP growth percent changes during the crisis episodes, averaged a severe drop of 6.1 percent. This compares with an average annual change of 314 percent for the non corporate crisis negative growth years, or 1 percent excluding the sharp early 1980s contraction of Poland. The magnitude of the contractions for the east Asian countries is broadly comparable to those of the earlier crisis episodes, taking into account the favorable external environment that benefited Mexico in 1995 (Roubini et al., 1998). Although the severity of these downturns reflects many factors not directly related to the corporate sector, such as banking sector problems, external shocks and rigid macroeconomic policies, the evidence presented below suggests the corporate sector plays a key role.
Industrial production also contracts sharply during a corporate crisis relative to the historical experience (Table 5, Figure 5). These declines are also very large based on the trough value compared to the pre-crisis average. During those episodes industrial production quickly dropped to levels well below trend. Indeed, shortfall of the level of industrial production from its trend is equivalent to several standard deviations for most of the episodes. Moreover, these shortfalls are in most cases the largest over the period of data availability. In some cases, the extent and timing of the downturn reflected the rapid tightening of corporate balance sheet constraints by reform, as was the case in Hungary following the introduction of a tough bankruptcy law (Cottarelli et al., 1999).
Comparison of the crises over time indicates that the speed of the contraction in industrial production may be accelerating. The number of months between the pre crisis peak of industrial production and its crisis trough fell from around 20 months during the crises of the early 1980s and 1990s to 13 months for Mexico in 1995 and further to 12 months on average for the east Asian countries (Table 5). This decline suggests that the speed of the transmission of the triggering shock to output has picked up over the past two decades possibly reflecting higher levels of unbalanced corporate leverage and capital and goods market integration.
1/ Data end in 1999 except for Poland (1995), and begin in 1975 except for Poland (1985). Hungary (1986), and Thailand (1995).
2/ Seasonally adjusted industrial production index less Hodrick-Prescott trend (smoothing parameter of 70,000).
The large contractions in economic activity correspond with sharp declines in investment (Table 6). Indeed, in six of the nine cases, the decline in aggregate real investment exceeded the overall GDP contraction. These investment declines were the largest over the period of data availability (generally 1975-99) for each of the nine countries examined here. The dominance of the investment declines, as opposed to the other components of demand, suggests that corporate distress played the key role in these large contractions. Moreover, the negative contribution of investment to growth is increasing over time (Table 6 and Figure 6).
projections for 1999.
Further evidence on the links between recession and corporate leverage is provided by descriptive cross section regressions of 1998 growth for twenty one emerging market countries. These regressions may provide some insight into understanding why only a few countries experienced severe recessions, even though all were hit by a cutoff of capital inflows.5 To understand this disparity, growth for 1998 for 21 countries were regressed on variables commonly cited as explanations for the Asian crisis (Table 7). The dependent variable is the difference between real growth in 1998 and trend growth during 1987-96. The independent variables predate the output contraction (except for the capital inflow and real interest rate measures) and therefore could be considered exogenous.