Exchange market pressure and currency crises in Turkey: an empirical investigation
thesisposted on 21.02.2011, 09:24 by Mete Feridun
This thesis investigates the determinants of exchange market pressure and currency crises in Turkey over the period 1989:09 and 2001:04 using three empirical methodologies: the Autoregressive Distributed Lag (ARDL) bounds testing approach to investigate the short-run and the long-run dynamics of exchange market pressure; the binary logit and ordered logit models in order to identify the determinants of currency crises; and third, it applies the signals approach to identify the leading indicators of currency crises. The findings of the thesis have indicated that speculative pressure in the foreign exchange market and currency crises in Turkey in the sample period cannot be attributed entirely to a single cause and that these two phenomena are a result of a diverse set of factors. The results also suggest that both speculative pressure in the exchange market and currency crises in Turkey are linked to the reversals in the capital flows. The findings have also indicated that another important factor which has given rise to speculative pressure in the exchange market and currency crises in Turkey is the weaknesses in the banking sector balance sheets, such as the overexposure to foreign exchange, liquidity and credit risks. The results of the thesis have also revealed that currency crises and speculative pressure in the foreign exchange market are linked to the overvaluation of the Turkish lira. Above all, the results have indicated that currency crises and speculative pressure in the foreign exchange market are not necessarily driven by common factors and that it is misleading to classify explanatory variables strictly as statistically significant and insignificant in the context of currency crises. Last but not least, the findings of this thesis have also suggests that statistically insignificant variables could still convey information regarding the imminence of currency crises if used in a non-parametric signals model.
- Business and Economics