Timing the Mexican 1994-95 financial crisis using a Markov switching approach
By Moritz Alberto Cruz | Fall 2006
Coauthor: Edmund Amann
From the Introduction: One of the most well-worn arguments within the vast and divergent literature concerning recent financial crises is that they stem from changes in market confidence. Such confidence is held to rest on so-called economic fundamentals and domestic political conditions. As soon as the fundamentals diverge from what may be considered sound, so it is argued, the confidence of investors ebbs away causing progressive withdrawals of capital overseas. In emerging economies, such withdrawals can have a devastating impact. Once private agents begin to withdraw capital abroad it consequently becomes increasinglydifficult to maintain steady growth. When the scale of the withdrawals becomes sufficiently large, in the absence of capital controls, contractions in GDP indubitably follow.
Within this setting, it is conventionally asserted that the start of any financial crisis is associated with the decision to float the exchange rate freely. The end of crisis occurs once the ensuing currency depreciation triggers export growth recovery. On a superficial level, there is some evidence to suggest that such an explanation fits the facts of a range of emerging market financial crises experienced over the past two decades. In the case of Brazil, for example, the decision to allow a float of the Real at the start of 1999 became associated with a sharp reduction in foreign capital inflows (Amann & Baer, 2003). The scale of this reduction was so accentuated that only under the auspices of an IMF standby facility was Brazil able to continue to meet its debt servicing obligations. Nevertheless, short run growth performance was very badly affected. At the same time devaluation was associated with severe trauma in the financial system despite the fact that currency floatation had been widely anticipated. Some 2-3 years after the crisis, however the economy had clearly entered a period of recovery, the most notable feature of which was a strong improvement in export performance. The rise in exports was closely tied to the now highly competitive valuation of the Real.
While explanations such as these appear superficially persuasive, the question arises as to whether meaningful conclusions can be drawn from an examination of the real objective facts of an event that has been associated with changes in market confidence, which itself a subjective rather than objective phenomenon. In other words, it is important to draw a methodological distinction between variations in the fundamentals and agents’ perceptions of them. One should not presume that alterations in the fundamentals always map in a direct and consistent manner into the evolution of market sentiment. Consequently we propose to adopt an alternative approach, strongly influenced by the seminal contributions of Minsky (1982, 1986). Minsky focuses his analysis on the relevance of expectations in the endogenous evolution of an economy. He argues that, in order to understand how and when financial crises start and end, we need to employ and analyse variables that allow us capture subjective changes in market confidence.
The Mexican financial crisis of 1994-95 provides an ideal setting for an exercise of this nature. The crisis, famously dubbed “the first of the 21st Century”, has some similarity with those of Argentina and Brazil in that it eventually entailed floatation and substantial currency depreciation. However, in some regards the Mexican crisis stands apart from its counterparts elsewhere in Latin America or, indeed, Asia. This is to do with the fact that its emergence was linked not to sharp deteriorations of the internal or even the external balance (two key fundamentals), but rather, to the growing exposure of private agents to risk based on their rapid accumulation of debt.
Download: Timing the Mexican 1994-95 financial crisis using a Markov switching approach (PDF, 185.45 K)blog comments powered by Disqus