policy whilst delivering short-term welfare gains would be outweighed by the long-run deadweight loss of social welfare due to the undermining of reputation
10
and loss of policy credibility. Second-generation
models have several noteworthy features. First, they are nonlinear, meaning that they allow policy-makers to react to the state of the economy, unlike the first-
11
generation models, which are state-invariant. Second, these models allow for multiple equilibria where maintaining the exchange-rate peg in the absence of shocks is the optimal outcome and de- valuing or abandoning the peg in the face of severe shocks can also be an equilibrium outcome. These second-generation models capture the time-incon- sistent policy behaviour that caused the crisis in the European Exchange Rate Mechanism (ERM) in 1992 and the financial meltdown in Mexico (the Tequila crisis of 1994).
The pre-crisis fundamentals for 1996 reported for Asia-5 (Table 3) reveal that these economies were experiencing strong growth rates, budget surplus to GDP ratios, moderate inflation rates of about 6%, high savings rates of over 32%, trade openness indicators of nearly 39% and credit ratings that were higher than investment grade. These stylised facts fail to support the first-generation models that attribute economic crises to bad macroeconomic fundamen- tals nor do they lend support to the second- generation models that attribute crises to the pursuit of time-inconsistent macroeconomic policies. Hence, the causes of the Asian economic crisis cannot be explained by these canonical models and newer explanations have to be sought.
Alternative explanations of financial crises in terms of rational self-fulfilling panic may fit the bill. In these newer panic models an investor in a project could be illiquid in the short term but could generate a cash flow and be solvent in the long term. However, the