| This dissertation presents a new way to study financial crises. The traditional approach uses macroeconomic statistics. While useful, this has serious limitations. By contrast, this dissertation reveals the personal story behind the big picture for one such case using a unique, fortuitously timed household survey of consumption expenditures collected throughout Turkey's 1994 Financial Crisis. The second chapter, following an introductory one, studies household welfare by analyzing nondurable goods spending during the crisis. A calibration analysis is developed to compare the observed risk-sharing with the model of perfect risk-sharing implicit in the representative agent models often used in crisis analyses. The chapter shows that risk-sharing was incomplete and socio-economic groups were impacted differently. Moreover, the welfare loss associated with a lack of risk-sharing was significant. The third chapter looks at the effects of crisis-induced employment uncertainty on durable goods purchases. Using Heckman's two-step procedure with chapter two's micro data, I find higher unemployment risk households to be less likely to buy durable goods, even after controlling for differences in incomes. The impact of uncertainty on the size of durables spending, however, is found to be less conclusive. The fourth chapter contextualizes the Turkish experience. Using aggregate data from sixteen crisis-hit countries, this chapter shows that overall consumption declines during crises, and durable goods spending is more sensitive and drops farther. These results mirror those found in the Turkish case, suggesting a general consistency in the pattern. |
