MỤC LỤC
- Identifying the crucial indicators that contribute to the EWS models to predict the currency crises in emerging markets.
Research questions
The scope of the thesis
The structure of the thesis
However, Berg and Pattillo (1999) argued that this definition could lead to false positives because some countries were regularly experiencing a fluctuation larger than 25% in exchange rate which were not be problems when taking the 25% in depreciation. They take “the real exchange rate and the interest rate are intended to account for differences in inflation rate across countries and over time” (Bussiere and Fratzscher, 2002, p.9).
Although, there were no specific events that might attribute to this generation of currency crises, the occurrences of currency crises following the 1997-98 Asian Financial Crisis such as Russia (1998), Turkey (2000-2001), and Argentina (2001-2002) have rose up the interested to find all possible causalities and linkages, which different from factors already known in three theoretical models of currency crises, could lead to a currency crises. Prudential supervision, accounting and disclosures requirements, legal and judicial systems, bureaucratic quality, government stability, absence of corruption, law and order, absence of external conflict, election, absence of internal conflict, exchange rate, capital control, absence of ethnic tensions, central bank independence, deposit insurance, financial liberalization and legal origin.
They identified many variables of macroeconomic fundamentals such as the growth of real exchange rate, the growth of broad money, domestic credit growth, current account surplus (or deficit)/ GDP ratio, reserve loss, export growth, import growth, Short-term debt/reserve and institution factors such as government stability, control of corruption, law and order, external conflict, internal conflict, voice and accountability, regulation quality. However, due to data’s limitation of institutions variables that available from 1996, this paper used the short period time (1996-2005) that did not cover all the crises occurring during 1990s such as the crises of Mexico, Turkey in 1994, even the shorten period before the Asian crises in 1997-1998 with the predicted time is 18 months.
Conceptual framework
After define the time when crisis occur, with the aim to predict the likelihood occurrence of currency crises or the currency crisis occurrence within the specific time horizon, this thesis transforms the contemporary currency crisis variables into a forward dependent variable Yi,t, which is defined as. It means that, when given the EWS model, alternative the value of parameters at time t, we get the probability of crisis occur within period t + k (k is the specific prediction time) by equation (3.1); then, we compare the probability of crisis with the cut-off points. At this time, there are two common methods to select the cut-off, this is given the cut-off in different level such as 0.5 or 0.25 even 0.1 and calculate the signal of crisis then they will choose which the best among them, or they supposed the loss function and qualitative for the policy marker so that the policy maker can decide the useful function by their own choice (Bussiere and Fratzscher, 2002).
On the contrary, the specification curve is up slope due to the higher cut-off point is the higher number of non-crisis signal issued, and higher the percentage of correctly non-crisis identified. Although, the shape of sensitivity and specificity curve is depended on the EWS specific model, the optimal cut-off always defined at the intersection of both sensitivity and specificity curve as Figure 3.2. Most of economic variables data taken from the International Financial Statistic (IFS, CD-ROM, 2011), some data taken from the World Bank (WB) and all the institution variables taken from the International Country Risk Guide (ICRG, 2012).
The minimum and maximum value of Short-term debt/reserve in whole time between 9.05% and 236.2% same as the value in the tranquil time, however, the minimum value in the crisis time is different (32.28%). For instance, with the index value from 0 to 12, the average value of government stability in the whole time, tranquil time or crisis time is around 8.2% - 8.8% and the average value of internal conflict is 8.5% - 9.6%. Although have the index value from 0 to 12, the external conflict has the average value is so low in whole time, tranquil time and crisis time, it is approximate 3.4%.
Besides that, the marginal effect of reserve loss is 5.55e-06, it means that when holding other variables at mean value, the reserve loss variable increases 1 unit, it increases the probability of currency crises by 0.0000055 units. Real exchange rate growth (RER): the marginal effect of this variable is 4.17e-06; it states that, when holding other variables constant, the growth of real exchange rate increases 1 unit, it will cause the probability of currency crises increase 0.00000417 units. However, Shimpalee and Breuer (2006) noted in their paper that this variable also possible to have the positive sign, because when the index value if government stability is high it may associate with the autocratic regimes and an absence of democracy and voice.
It stated that, ceteris paribus, if the index value of internal conflict increases by 1 unit, the probability of currency crises will increase by 0.0007755 units. However, in the case the policy-maker is cautions, they define the cost of missing alarm is higher than the false alarm, or they stated the cost for recovery crisis when the crisis actually occurred is higher than the cost for preventing the crisis occur. Besides that, with the high false alarm increase, the policy-maker face with the risk of lost cost for pre-emptive actions such as implement some macroeconomic policies prematurely in order to prevent the currency crises that actually were not occur (Comelli, 2014).
Compare with the optimal cut-off at 13.27%, it is easy to conclude that this EWS model issued signals of crises that will occur in Indonesia, Malaysia, Philippines and Thailand within the year 1997 and had no signal issued in Turkey, it stated that Turkey would not have to face with crisis in 1997. Consequence, the crisis already happened, it started from Thailand in July 1997, and spread to rest of Asia countries such as: Malaysia (July, 1997), Indonesia (August, 1997) and Philippines (October, 1997) and Turkey was spending tranquil period during this time. The false signals of prior crisis occurrence could be explained that, when it issued the signals of crises occurring, the government of these countries, thus, had some policies to prevent the crises occurring such as sold the foreign reserves, or increase the interest rate to keep the peg exchange rate or keep the currency not devaluated.
Besides that, among macroeconomic variables, the reserve loss, short-term debt, current account/GDP and the growth real exchange rate still have the correct sign and significant at 1% level while the export, import change the sign, and import and GDP growth lost their significant. Moreover, the government stability change to positive sign and significant at 1%, it meant that, in this area, the stability of government is very crucial and have the different effects on currency crises from Asian areas. With results of robustness test, it stated that with different areas, although, it has the difference EWS model performances, this model can forecast correctly most of currency crises that occurred in the past with these indicators and logit method as well as suitable cut-off point methods.
Short-term debt/reserve, even though, is not significant in Tuan & Hoai (2009) when they combine it with macroeconomic variables, it is a good indicator in our model and Leblang and Satyanath (2008) and also significant in Bussiere and Fratzscher (2002). As you can see, the institutions variables just mentioned in the recent studies, among those variables, government stability is always significant when used to predict the currency crisis. Besides that, while Tuan (2009) is not found the significant of Law and order, our model and Shimpalee and Breuer (2006) found it significance.
Firstly, due to lacking of data source, this study employed a limited number of variables, some importance indicators related to global factor such as US interest rate, growth of world oil prices could be considered as sound fundamental indicator for extended version of this model besides other institutional variables that may be relevant such as bureaucratic quality, deposit insurance, capital control and financial liberalization. Even though logit regression is the main technique used for most of EWS model, it has some limitations of its own: it is not possible to examine which variable is the most effect on the currency crises and this model is not good method to measure the interaction between explanation variables. In addition, the next researches could also apply another technique such as Artificial Neural networks, Markov Switching, Ordinary Least Squares (OLS) or whatever methods to construct an EWS model that could measure the effect of each variables and find out which is the most effective indicator so that the model may use fewer indicators but it is the best indicators to observe.
Kaminsky, Lizondo and
IMF (1998) 6. Esquivel and Larrain (1998)
Berg and Pattillo (1999a) 8. Berg and Pattillo (1999b)
Glick and Moreno (1999) 10. Vlaar (2000)
Kamin Schindler and Samuel (2001)
Bukart and Coudert (2002) 14. Kumar, Moorthy and Perraudin (2003)
Edison (2003) 16. Peltonen (2006)
The first column presented the result of Model 1, it stated that all macroeconomic variables have the significant at 1% level, except export growth at 5% level and real exchange rate has not significant. The Table E.2 stated LR chi2 = 242.72 and significant, it meant that it fail to reject the null so it is should not to drop the institution variables, in another way, the Model 2 is better than Model 1. Besides that, in the Table E.3, the significant of _hat (p = 0.000) showed that the Model 1 included the meaningful indicators while _hatsq also significant stated that this model has the specification error.