Price Index and Trading Volume: Which one is the best to predict financial crisis? The case of some Asian and Latin American Markets Dominique Gu´egan∗, Octavie Mbiakoup†
Abstract The objective of this paper is to use both volatility and volume as possible leading indicators to predict financial crisis. By using exclusively information ex-ante , we can evaluate the behavior of these two variables before some financial crisis in some emerging markets. This paper suggests a model in which the thresholds of the degradation of price and the growth of the volume reach a certain level whereby we can predict market crisis. This model can be used to explain the crisis occurred in East Asia and Latin America between 1994 and 2002. Due to the fact that this model only predicts a small quantity of false crisis, we can conclude that price and volume, can be used to successfully predict a financial crisis as turning point in financial market. Keywords: Volatility, Volume, Turning Point, Predict, Financial Crisis. JEL classification: C45, C53, G11
∗
ENS Cachan D´epartement Economie et Gestion, MORA-IDHE and Senior Academic Fellow de l’IEF, 61, Avenue du Pr´esident Wilson, 94230 Cachan, e-mail:
[email protected] † ENS Cachan, MORA-IDHE, 61, Avenue du Pr´esident Wilson, 94230 Cachan, email:
[email protected], www.idhe.ens-cachan.fr/MORA
1
1
Introduction
Recent years, we witnessed an increase in currency crisis affecting a large number of emerging countries. Some similarities have been observed in the way these crises unfolded namely: a loss of foreign exchange reserves, capital outflow and more often a sudden depreciation of the currency. If similarities between these economies also exist before currency crises occurred, they could be used to predict these crises. Indeed, each crisis is specific: the debt crisis in Latin America in the 1990s is different from the Mexican crisis in 1994, which differs from the crises of Southeast Asia in 1997. More over, within each of these episodes, each crisis presents different specificities. However, some common features emerge. Theoretical studies have identified several type of crises (for a survey, see Jeanne, 1999). Theoretical work provides some guidance when choosing potential leading indicators, which should reflect fundamentals as well as any variables able to influence the market expectations. However, it neither allows discrimination between competing indicators, nor does it enable their respective weights to be determined. This is the reason why an important literature has been developed. For an overview, see Kaminsky et Reinhart (1997), Frankel and Rose (1996), Goldstein et al (2000), Burkart and Coudert (2002). All those studies can be classified, according to different criteria : the country analysed, the periodicity of the time series, the methods implemented and most importantly, the choice of the time series. More often, the time series are related to classical fundamentals of economy such as GDP, exchange rate, stock return and deposit of currency ratio,etc.Only a few studies used price index as leading indicators, while, to our knowledge, none used trading volume. So, when an economy is slack(does not walk well), this may be noticed in some financial data (financial market). In this case, assets loose their value (decrease of the increase price) and each one wanted to sell it (increase of the trading volume). It was the case in US stock market in 1987, the Dow Jones lost 12% of its value while the trading volume rose up to 77%. Many research have come to the conclusion that asset prices have a significant predicting power on future activities. In fact, for McGuckin et al (2003), Bellone, Gauthier and Le Coent (2005), in the United States, used a composite index as the main leading indicators provides a quite accurate forecast of the business cycle with 3 to 6 months predicting horizon. Anas and Ferrara (2004) study the case of industrially countries and reach to the same conclusion. Between October 14 and October 19, 1929, major indices of market valuation in the US dropped 30% or more, while trading volume attains 9 millions in one day. In October 19, 1987, a date that subsquently became known as black monday, the Dow Jones plummed 508 points, loosing 22.6% of its value. The S&P dropped 20.4%, falling from 282.7 to 2
225.06. At the same moment the exchange volume rose to 77%. Between October 20 and October 23, 1997, Hang Seng Index dipped by 23%. In July 1997, the PSE (Philippine Stock Exchange) Composite Index fell to some 1000 points from a high of some 3000 points. The Thai Stock Market dropped 75 % in July 1997. The same phenomena have been observed in many emerging countries during each episode of financial crisis. Basing on these observations, this paper aims as constructing a leading indicator of financial crisis for a set of 12 emerging countries in the period 1994-2002 in order to predict financial crises. The rest of the paper is organised as follows. Section 2 describes an empirical literature of some indicators and crises. After recalling the different existing studies on the topic, we give a proper definition of turning point relating to financial crises, justify the possibility to use both index price and exchange volume as leading indicators, and construct the leading indicator in section 3. We apply our model in a set of twelve emerging markets and present the results in section 4 with some forecasting hypotheses. Section 5 concludes the paper with some remarks and gives some indications for future research.
2
Empirical Literature
This section begins with a description of the various methodologies and variables that have been used in the empirical literature to characterize the period preceding a crisis and to assess the probability of such crisis. It then proceeds to narrow the list of potential leading indicators to those variables which appear to have worked best. Table 7 in annexe summarizes some selected empirical studies on financial crises. The first column lists the study, the second describes the sample periods and the periodicity of the data and the third provides information on the countries covered and the type of episode examined. The fourth column lists the economic variables that have been used as indicators, and the last column sketches certain features of the methodology used. The authors examine sample that runs from the early 1950 to mid 1995 and cover both industrial and developing countries. Most of the paper examine the experience of various countries and study several crisis episodes; only a few papers focus on a single country. The studies also vary with respect to how a crisis is defined. Most of the studies focus exclusively on devaluation episodes. Some of them examine large and infrequent devaluations, while others include in their sample small and frequent devaluations. A few study adopt a broader definition of crises. In addition to devaluations, they include episodes of unsuccessful speculative attacks (attacks that were averted without a devaluation), but at the cost of a large increase in domestic interest rate. Regarding the methodologies used, the various papers can be grouped into four broad categories. A first group of papers provides only a qualitative discussion of the cause and develop3
ments leading to the currency crises(for instance, Dornbush, Goldfan, and Valdes(1995) stress an overvalued exchange rate). A second group of papers examines the stylized facts of the period leading up to and immediately following the financial crisis. Sometimes, the pre-crises periods behavior of a variable is compared to its behavior during tranquil or non-crisis periods for the same group of countries(for example, Eichengreen, Rose and Wyplosz(1995), Frankel and Rose (1996)). In other instances, the control group is composed of countries where no crisis occurred. A third group of papers estimate the probability of devaluation one or several periods ahead, usually on the basis of an explicit theoretical model. These papers include individual and multi-country panel studies (Frankel and Rose (1996). In a related spirit, Sachs, Tornell and Velasco (1996) seek to identify those macroeconomic variables that can help explain which countries were vulnerable to contagion effects following the Mexican crisis of December 1994. A fourth type of methodology is used by Kaminsky and Reinhart (1996). Their article presents a nonparametric approach to evaluate the usefulness of several variable in signaling an impending crisis. It can be interpreted as an extension of the methodology that compares the behavior of variables in periods preceding crisis with that in a control group. This approach involves monitoring the evolution of a number of economic variables whose behavior usually departs from normal in the period preceding a financial crisis. Deviation of these variables from their normal levels beyond a certain threshold value, are taken as warning signals of a financial crisis within a specified period of time. Based on the track record of the various indicators, it is possible to assess their individual and combined ability to predict crisis. In this paper, we use the same approach as Kaminsky and Reinhart. The main difference is the use of only two variables. These variables are price index and exchange volume in financial markets. This approach is explained in detail in the next section.
3
Methodology
3.1
Definition of a Turning Point
In time series analysis, the word ”Turning Point” is used when a series, which had been increasing (decreasing) reverses and, for a time, decreases (increases). In other words, a turning point date as occurring when a swing in one direction ends and a swing in the other direction begins.Considering this definition, the emerging markets experienced many1 crisis for the 1994-2002 period. The starting dates for these events vary depending 1
Michel Aglietta, Macro´economie financi`ere Tome 2, 1995-2005
4
on the source:
1. Mexico (Tequila effect, December 20, 1994) 2. Bresilia, March 1995, 13 January 1999 and October 2002 3. Thailand July 2, 1997 4. Philippines July 97 5. Malaysia, July 1997 6. Indonesia August 1997 7. Hongkong October 22, 1997 8. Korea December, 1997 9. Russia 1998 10. Turkey November 20, 2000 11. Argentina July 2001 and June 2002
3.2
Choice of the series
The studies reviewed in this paper used, contrary to other study, only two indicators: daily price (stock market index or shares price) and exchange volume (turnover activity). 3.2.1
Daily Data: Why?
This work is based on daily data, whereas the bulk of the studies carried out so far uses annual , quarterly or monthly data. Although these types of periodicities allows easier access to data, it cannot really take into account financial phenomena, such as change in stock market, which occur with very high speed, especially in periods of crisis. Furthermore, annual time series weakens the predictive value of leading indicators because it is unclear whether a crisis took place at the beginning of the year or at the end. Several authors following Kaminsky-Lizondo-Reinhart have worked with monthly data. In this case too, as with quarterly data, it would be difficult to know exactly when the crisis started (end of the month or middle of the quarter). By using daily data, we are able to overcome some of theses difficulties, and also to consider questions, which would be out of reach with annual data. For example, structural indicators may be more important for an horizon in excess of one year and financial-flow indicators may become increasingly important as the predictive horizon shrinks (6 or 3 months).
5
However, it must be pointed out that the use of daily data adds some noise and makes thus the signal extraction more complex. This problem is resolved by using a set of growth rate, so as to choose the one which give a good signal. Many variables are not available on a daily basis. This justifies too the used of only two variables.
3.2.2
Volume and Price: How?
In the literature, dating crisis or predicting crisis is usually done by using classical fundamentals of economy. Only a few papers include composite price index and none takes trading volume into consideration. In the financial markets, rising share prices, for instance, tend to be associated with increased business investment and vice versa. Share prices also affect the wealth of households and their consumption. Therefore, central bank tend to keep an eye on the share and control the behavior of the stock market and, in general, on the smooth operation of financial system functions. Financial stability is the raison d’ˆetre of central banks, along with inflation stability. Further to the increase of interest rates, at the first stagnation of currency, the paying off of interests exceeds stock market profits. Consequently, many investors are compelled to sell their share to cover loans, which causes currency to decline and unleashes a chain reaction. By using information ex-ante, it appears that before each financial crisis mentioned in the literature, the trading volume, as well as the price index, present some patterns, which disappear after the crisis. This leads us to think about those two variables as possible leading indicators. But, considering that the market collapse the day of crisis is important only if we want to date a financial crisis. The aim of our paper is not only to date, but also to predict the occurrence of a financial crisis. In all countries, we have observed a rise of exchange volume and a drop of price index, in the same period. Because the graphical approach has disadvantages(sometimes, a drop of price index or a rise of exchange volume is not a sign of financial crisis), the results in the figure are essentially hypothesized. Figure 1 and 2 represents price index and exchange volume in some countries, surrounding the date of the financial crisis in each country. The common features we have already mentioned appears in those figures. When looking for Mexico time series, the daily price index (Figure 1, (d)) collapsed on December 1994, while the trading volume (Figure 2, (d)) rise the top, around the same day. Else, in the case of Thailand (Figure 1 and 2, (a)),the price failed in July 1997 and at the end of the year, while the trading volume attains its highest value from the beginning of the year to July 16, two weeks after the collapse of the index price. In the case of Hongkong, the same conclusion are reach (Figure 1 and 2, (b)), and the crisis occurred in October 20, 1997.
6
(a)
(b)
800
2500
700 600
2000
500 1500
400 300
0
50
100
150
200
250
0
50
100
(c)
150
200
250
(d)
160 140
2600
120
2400
100 2200
80 60
0
50
100
6
150
200
2000
250
0
20
40
60
6
(e)
x 10
80
100
120
140
(f)
x 10
5 3 4 2 3 0
50
100
150
200
1
250
0
50
100
150
200
250
Figure 1: Index Price for some markets surrounding their financial crisis: Thailand in 1997(a), Hongkong in 1997(b), Korea in 1997(c), Mexico in 1994(d), Argentina in 2000(e) and Argentina in 2001(f).
4
6
(a)
x 10
(b)
x 10 2.5
10
2 1.5
5
1 0.5
0
0
50
100
4
150
200
0
250
2
2
1
0
50
100
4
150
200
0
250
100
0
150
200
250
(d)
20
40
60
4
(e)
x 10
50
x 10
4
0
0 5
(c)
x 10
80
100
120
140
(f)
x 10 3
2
2 1 0
1 0
50
100
150
200
0
250
0
50
100
150
200
250
Figure 2: Exchange volume for some markets surrounding their financial crisis: Thailand in 1997(a), Hongkong in 1997(b), Korea in 1997(c), Mexico in 1994(d), Argentina in 2000(e) and Argentina in 2001(f) .
For Argentina, (Figure 1 and 2 (f)) suppose two crisis: in July 2001 and in December 2001. We remark that a turning point is not necessary the moment of crisis and the construction 7
of the leading indicator will take into consideration this situation .
3.3
Construction of leading indicator
The potential leading variable is evaluated by examining the past of the series before the collapse. Five emerging countries for six different crises are used as learning set: Thailand, Hongkong, Korea, Mexico, and Argentina. All those countries experimented a financial crisis. The plot of the series in the crisis’s year as we have done at the beginning has permitted only to observe that at the moment the crisis occurs, we have a rise in exchange volume and a collapse in price index. But because the series are too volatile (due to daily frequency), we construct many series of returns for different lag p (p=1,2,5,10,15,20,30,40,50). The idea is to obtain a lag surrounding the day of the crisis. We define the growth rate over p days by the following formula:
Yt =
Xt − Xt−p ∗ 100 Xt−p
(1)
where Xt is the index price or Exchange volume and Yt is the p-days return. Notice that for p=10, Yt is the return over 10 days, (ten trade’s days). We plot each Yt . For the case of p= 1,2,5,10 the series of price index is still too volatile and the signal is difficult to extract. On the contrary, in the case of 50 (Two months and a half)for price index and 15 till 50 for exchange volume, the series is too smooth and the day of the crisis disappears. So, we consider 4 returns: 15, 20, 30, 40 for price and 1, 2, 5 and 10 for exchange volume. The objective is to choose the return which gives the suitable information. This return should be common to all the markets, and can differ for each variable.
3.3.1
Asian Markets
• Thailand The crisis occurred in Thailand in July 2, 1997. In Figure 3, only the case p=30 give a date surrounding the crisis date. We summary the results in the following table: Table 1 clearly show that the best growth rate is with lag p=30, only two days far from p 10 15 20 30 40 50 Date 17/06 08/07 24/06 30/06 24/06 17/06 Table 1: Crisis day for different lag the official date of the crisis. • Hongkong 8
p=10
p=15
60
60
40
40
20
20
0
0
−20
−20
−40
0
50
100
150
200
−40
250
0
60
120
p=20 60
60
40
40
20
20
0
0
−20
−20
−40
0
40
80
120
180
240
p=30
160
200
−40
240
0
46
92
p=40
138
184
230
p=50
60 20
40 20
0
0
−20
−20 −40
0
40
80
120
160
−40
200220
0
70
140
210
Figure 3: Thailand p 10 15 20 30 40 50 Date 28/10 28/10 28/10 28/10 27/10 26/10 Table 2: Crisis day for different lag in Hongkong In Hongkong, the crisis took place in October 22, 1997. Figure 4 represent price index with different lags, and Table 2 give the date of the crisis for each lag. In opposite of the case of Thailand, here, we are able to choose any lag. So for convenient, we will consider p=30. • Korea In Korea, the crisis occurred in December 1997. When p=30 in this case too, the date of p 10 15 20 30 40 50 Date 28/10 04/11 28/10 02/12 27/10 26/10 Table 3: Crisis day for different lag in Hongkong the collapse is not too far from the exact date of the crisis. 3.3.2
Latin American Markets
We proceed as in Asian markets. • Mexico
9
p=10
p=15
20
20
0
0
−20
−20
−40
0
50
100
150
200
−40
250
0
50
100
p=20
150
200
250
p=30
20 20 0
0
−20 −40
−20 0
40
80
120
160
200
−40
240
0
40
80
p=40
120
160
200
240
p=50 40
20
20 0
0
−20
−20 −40
−40 0
44
88
132
176
−60
220
0
44
88
132
176
220
Figure 4: Hongkong
p=10
p=15
20
20
10
10 0
0
−10
−10 −20
−20 0
60
120
180
−30
240
0
60
p=20
120
180
240
180
240
180
240
p=30
20 20 0
0 −20
−20 0
60
120
180
−40
240
0
60
p=40 20
20
0
0
−20
−20
−40
120
p=50
0
60
120
180
−40
240
0
60
120
Figure 5: Korea
The crisis occurred in Mexico in December 1994. For p=15, the date surrounding this crisis is pointed out. • Argentina Argentina have experimented two financial crisis: July 2001 and June 2002. In the two cases (Figure 7 and 8), the date of the crisis is pointed out for p=15. 10
p=10
p=15
15 10
10 5
0
0 −5 −10
−10 0
30
60
90
120
0
30
p=20
60
90
120
p=30
20 20 10 0
0 −10
0
20
40
60
80
−20
100
0
20
p=40 20
0
0
0
20
40
60
60
80
100
80
100
200
250
p=50
20
−20
40
80
−20
100
0
20
40
60
Figure 6: Mexico94
p=10
p=15
30
40
20
20
10 0
0 −10 −20
−20 0
50
100
150
200
250
0
50
p=20 40
20
20
0
0
−20
−20 60
120
180
240
0
60
p=40 40
20
20
0
0
−20
−20 44
88
132
120
180
240
p=50
40
0
150
p=30
40
0
100
176
220
0
44
88
132
176
220
Figure 7: Argentina2001
Finally, we consider Equation 1 with p=30 for East Asian Markets, and p=15 for Latin American Markets. The same work is done with exchange volume (See the graphs in Appendix). The main difference with price is that it is not necessary to distinguish between Asian and Latin American Markets. In fact, for all the countries, a return over one day is sufficient to point out the time surrounding the crisis.
11
p=10
p=15 60
40
40
20
20
0
0
−20
−20
0
50
100
150
200
250
0
50
p=20 60
40
40
20
20
0
0
−20
−20
60
120
180
240
0
60
p=40 60
40
40
20
20
0
0
−20
−20 0
44
88
132
200
250
120
180
240
p=50
60
−40
150
p=30
60
0
100
176
−40
220
0
44
88
132
176
220
Figure 8: Argentina2002
3.3.3
The signals Approach
As mentioned above, this approach involves monitoring the evolution of two financial variables. When one of these variables deviates from its normal level beyond a certain threshold value, it is taken as a warning signal about a possible financial crisis within a specified period of time. For each country, crises are identified as a drop in price index and a rise in exchange volume. In the empirical application, mostly the authors used 10% threshold levels to detect the crisis. In this study, we showed that the -10% threshold level is not good enough to detect the emerging markets crisis. Here, threshold levels are chosen so as to strike a balance between the risks of having many false signals (which would happen if a signal is issued at the slightest possibility of a crisis) and the risk of missing many crisis (which would happen if the signal is issued only when the evidence is overwhelming). For each of the indicators, the following procedure is used to obtain the optimal thresholds that were employed in the empirical application. Here we consider only the case of price index: • For each country, we select the p for which the date (or around) of the crisis is pointed out. The final p should be common to all the countries or group of countries. • We plot the new series in the crisis’ year and observe it past before the crisis. • In the past of the series, we reper all the troughs. Each trough is a potential threshold.
12
• Finally, for each trough, we evaluate the delay for the trough of the crisis. We give, in Table 4, for some thresholds, the delay before the crisis. Thresholds Thailand Hongkong Korea Mexico Argentine1 Argentine2
-10% 35 /4crises 20 /2crises 25 /3crises 12 /1 crisis 15 /2 crises 15 /3 crises
-15% 30 /4crises 15 /1crisis 15 /1crisis 10 /1 crisis 10 /2 crises 10 /2 crises
-20% 23 /3crises 10 /1crisis 15 /1crisis Crise Crise Crise
Table 4: Thresholds and delays before a crisis for different markets, using price index In this Table, when the threshold reaches −10%, the crisis occurred in Latin American markets three weeks later. But in Asian markets, with the same threshold, it takes one month. When the threshold decreases, the delay decreases too and when it attains -20%, in Latin American markets, it is a crisis, while crisis is imminent (2 weeks) in East Asian markets. The number of potential crisis is given too in the same Table. It appears that when choosing a threshold equal to −20%, except the case of Thailand, we have only one crisis as seen in the literature in the two other countries: Hongkong and Korea. But the delay is too short to prevent the crisis, only three weeks. The same procedure is applied in volume. Figures are plotted in Annexe. We summarize the result in the Table below: Thresholds Thailand(40 days) Hongkong (35 days) Korea (10) Mexico (10) Argentine1 (10) Argentine2 (10)
100% 2 crises 3 crises 1 crisis 2 crisis 3 crisis 2 crisis
200% 1 crisis 3 crisis 1 crisis 2 crisis 3 crisis 1 crisis
300% crisis 2 crisis crisis 2 crisis 2 crisis crisis
Table 5: Thresholds and delays before a crisis for different markets using only volume In this table, the number behind the countries in brackets represent the number of days before crisis. The growth rate of volume attains a peak thirty days before the crisis in the East Asian Markets and 10 days in Latin American Markets. When using volume, it is easy to see that it helps to reduce the number of false crisis. For example, in Thailand, the use of the index price detect four financial crisis in 1997 while exchange volume detect only one.
13
4 4.1
Applications Data and data sets
For the construction of the leading indicator, we consider only the crisis’year and a learning set of five countries. We turn now to an application in many other markets.In this section we work with the returns of the twelve emerging markets. The sample includes data from 12 emerging countries: Argentina, Indonesia, Thailand, Korea, Chile, Malaysia , Philippine, South Africa, Taiwan, Mexico, Hongkong, Turkey. The number of countries is small in comparison to other studies that mix different types of countries: developed, emerging or developing. Like other recent studies, we have preferred to focus on emerging countries where a currency crisis could have potentially strong effects. The data set runs from 01/07/1994 to 31/12/2002. The starting date is conditioned by the availability of data in Mexico. The whole series contain 2218 returns.
4.2
Results
We consider a threshold for −20% in Asian Markets and −15% in Latin American Markets for price index. For exchange volume, in Asian and Latin American Markets, the better threshold is 200%, and the indicator detected the crisis one month before it occurred in Asia and two weeks in Latin America. For the price index, the indicator detected the crisis three weeks before it occurred in Asia and two weeks in Latin America. So, in Asia, volume signaled the crisis before the price index, but in Latin American we have a coincident signal. As appeared in Table 6, contrary to the learning set, volume gives more crisis than price index.
5
Conclusion
The aim of this paper was to see the possibility to use price index and exchange volume as leading indicators. We argued that the similarities in both trading volume and price index before each crisis could help to predict a financial crisis less than six months before its occurrence. The easiest model we have proposed works well for all the countries during the 1990s financial crisis using price index. However, we believe our work leaves two important questions unanswered that warrant further research. The first question is, can the knowledge of only those two variables be sufficient in practice to improve standard economic forecasts? Our analysis suggests that it may be fruitful to put together financial and economic fundamental so as to predict not only the major financial crisis, but also those caused by contagion effects.
14
Number of crisis by using Thailand
Price index 3
Hongkong
2
Korea Indonesia Malaysia
1 1 2
Turkey
3
Taiwan Chile South Africa Philippines Mexico
1 0 1 1 2
Argentine
5
Exchange Volume Date of crisis Many 09/02/97 30/06/97 15/12/97 2 28/10/97 09/07/98 Many 02/12/97 Many 04/09/97 Many 13/11/97 20/08/97 4 17/04/97 17/09/98 07/12/00 4 29/09/00 Many 8 27/08/98 Many 22/08/97 Many 15/12/94 27/08/98 Many 16/01/94 23/10/97 25/02/99 20/09/01 30/05/02
Table 6: Crisis in some emerging markets using the model The second question is , are the processes for turning points( in our context) of different important variables related? Our analysis suggests that sometimes there is a lag between the collapse of the price index and the increase of exchange volume. It would be worthwhile to explore, for instance, the possibility to model both exchange volume and price index as a joint process. That is far different from analysing correlations at high frequencies between the two variables, as is commonly done in the literature(Staiger, Stock and Watson, 1989). Therefore, those two variables can be used to measure contagion between markets and this should be done in the next paper.
15
6
Annexes p=1
p=2
300 200
200
100
100
0
0 −100
0
100
200
−100
260
0
100
p=5
200
300
200
300
p=10
400
400
200
200
0
0 0
100
200
300
0
100
Figure 9: Volume Exchange in Korea Market in 1997
p=1
p=2
200
200
150 100
100 50
0
0 −50
0
50
100
150
200
−100
250
0
50
p=5
100
150
200
250
200
250
p=10 400
200 200
100 0 −100
0 0
50
100
150
200
250
0
50
100
150
Figure 10: Volume Exchange in Thailand Market in 1997
16
p=1
p=2
1500
2000
1000
1500 1000
500
500
0
0
−500
−500
0
50
100
150
200
250
0
50
100
p=5
150
200
250
200
250
p=10
2000 2000 1000
1000
0
0 0
50
100
150
200
250
0
50
100
150
Figure 11: Volume Exchange in Hongkong Market in 1997
p=1
p=2 400
400
300 200
200
100 0
0 0
50
100
150
200
−100
250
0
50
p=5
100
150
200
250
p=10
600
800 600
400
400
200
200 0
0 0
50
100
150
200
−200
250
0
50
100
150
200
250
Figure 12: Volume Exchange in Argentina Market in 2001
17
p=1
p=2 400
200
300
100
200 100
0 −100
0 0
50
100
150
200
−100
250
0
50
100
p=5 300
300
200
200
100
100
0
0
−100
−100
0
50
100
150
150
200
250
200
250
p=10
200
250
0
50
100
150
Figure 13: Volume Exchange in Argentina Market in 2002
p=1
p=2
4000
1000
3000 2000
500
1000 0 −1000
0 0
50
100
0
50
p=5
100
p=10
600
800 600
400
400
200
200 0
0 0
50
−200
100
0
50
100
Figure 14: Volume Exchange in Mexico Market in 1994
18
Malaysia 60 40 20 0 −20 −40
0
500
1000
1500
2000
2500
3000
2000
2500
3000
2000
2500
3000
2000
2500
3000
Turkey 100 90 70 50 30 10 −10 −30 −50
0
500
1000
1500
Philippines 40 20 0 −20 −40
0
500
1000
1500
Taiwan 60 40 20 0 −20 −40
0
500
1000
1500
Figure 15: 30-days return of Index Price for some countries in the full period
Thailand 30 10 −10 −30 0
500
1000
1500
2000
2500
3000
2000
2500
3000
2000
2500
3000
2000
2500
3000
Korea 30 10 −10 −30 0
500
1000
1500
Hongkong 30 10 −10 −30 0
500
1000
1500
Indonesia 30 10 −10 −30 0
500
1000
1500
Figure 16: 30-days return of Index Price for some countries in the full period
19
Argentina 60 40 20 0 −20 −40
0
500
1000
1500
2000
2500
3000
2000
2500
3000
2000
2500
3000
2000
2500
3000
Mexico 40 20 0 −20 −40
0
500
1000
1500
Chile 40 20 0 −20 −40
0
500
1000
1500
South Africa 40 20 0 −20 −40
0
500
1000
1500
Figure 17: 30-days return of Index Price for some countries in the full period
Turkey 400 300 200 100 0
0
500
1000
1500
2000
2500
3000
2000
2500
3000
2000
2500
3000
2000
2500
3000
South Africa 2000 1000 0
0
500
1000
1500
Taiwan 500
0
0
500
1000
1500
Philippines 500
0
0
500
1000
1500
Figure 18: Return of Exchange Volume for some countries in the full period
20
Hongkong 1000 500 0
0
500
1000
1500
2000
2500
3000
2000
2500
3000
2000
2500
3000
2000
2500
3000
Korea 500
0
0
500
1000
1500
Thailand 500 400 300 200 100 0
0
500
1000
1500
Indonesia 1000 500 0
0
500
1000
1500
Figure 19: Return of Exchange Volume for some countries in the full period
21
22 1985-1995, monthly and annual
Sachs, Tornell and Velasco(1995)
20 emerging market countries
Indonesia, Japan, Malaysia, Philippines, Singapore, Korea, Thailand
20 countries, 5 industrial and 15 developing
107 devaluations with the respect to the U.S. dollar
Argentina, Brazil, Chile and Mexico
Country Coverage 105 developping countries
The real exhange rate, current account/GDP, M2/International reserves, saving/GDP
export/import, inflation, changes in net foreign assets, output gap
export growth, import growth, real interest rate, change in stock price, changes in reserve
import growth, changes in reserves, inflation, the real exchange rate
international interest rates, real exchange rate, short term borrowing
Indicators credit growth, external debt, current account/GDP
Table 7: Indicators of Crises: a Review of the Literature
1980-1994, monthly and quaterly
Moreno(1995)
1953-1983, annual
Kamin(1988)
1970-1995, monthly
annual and monthly
Golstein (1996)
Kaminsky and Reinhart(1996)
Sample and frequency 1071-1992, annual
Study Frankel and Rose(1996)
The emphasis is on explaining why some countries were more affected by the Mexican crisis than others
The emphasis is on testing whether the behavior of macroeconomic variables differs between tranquil and speculative periods
The behavior is examine 18 months before and after the crises and compared to the behavior of the same indicators during tranquil periods
The evolution of the indicators is examined 3 years before and 4 years after the devaluations and compared with the evolution of the same variables in the control group
No formal test, but why some countries were more vulnerable than others
Comments The evolution of these indicators around crises is compared to the behavior during tranquil periods
References [1] Ciccarelli. M: Measuring Contagion with Bayesian Time Varying Model, International Monatary Fund June 2003 [2] Giancarlo . Corsetti: Correlation analysis of financial contagion: what one should know before running a test TEX, Yale University, Univesity of RomeIII, and CEPR [3] Kamisky, Graciela L., and Carmen M. Reinhart, 2000. : On crisis, contagion, and confusion, Journal of International Economics 51, 145-168. [4] Burkart Olivier, Coudert Virginie: Leading indicators of currency crises for emerging countries , Emerging Markets Review 3 (2002) 107-133. [5] David A Umstead: Forecasting Stock Market Prices , The Journal of Finance, Vol.32, N◦ 2, September 16-18, 1976 (May, 1977), 427-441. [6] William E. Wecker: Leading indicators of currency crises for emerging countries Predicting the Turning Points of a Time Series, The Journal of Business, Vol 52, N◦ 1, Jan 1979, 35-50. [7] Kaminsky G., Reinhart C., (1996): The Twin Crises: the causes of Banking and Balance of Payments Problems International Finance Discussion Papers, Board of Governors of the Federal Reserve System, n◦ 544. [8] Kaminsky G., Lizondo S., Reinhart C., (1997): Leading indicators of currency crises, The World Bank, Working Paper. [9] Andersson E., Bock D. and Frisen M (2004): Detection of Turning Points in Business Cycles, ournal of Business Cycle Measurement and Anlysis, Vol 1, N◦ 1 [10] Adil Naamane: Les indicateurs d’alerte des crises financi`eres , Work in progress. [11] Anas J., Billio M., Ferrara L., Lo Duca M. (2004): A Turning Point Chronology for the Euro-zone, GRETA,Working Paper n.04.01. [12] Canova F (1994): Were Financial Crisis Predictable?, Journal of Money, Credit and Banking, Vol.26, N◦ .1, 102-124 [13] Chin D., Geweke J., Miller P. (2000): Predicting Turning Points, Technical Paper Series, Congressional Budget Office, Washington, DC. [14] Dornbusch, Rudiger, Ilan Goldfajn and Valdes (1995): Currency Crises and Collapses, Brookings Papers on Economic Activity, N◦ 2, pp. 219-95. [15] Frankel, Jeff and Andrew Rose(1996): Currency Crashes in Emerging Markets: An empirical Treatment, International Finance Discussion Paper N◦ .534 (Washington: Board of Governors of the Federal Reserve). [16] Goldstein, Morris(1996): Presumptive Indicators/Early Warning Signals of Vulnerability to Financial Crises in Emerging Market Economies, (Unpublished; Washington: Institute for International Economics) [17] Kamin, Steven B.(1988): Devaluation, External Balance, and Macroeconomic Performance: A look at the numbers, Studies in International Finance, N◦ 62, Princeton University, Department of Economics, International Finance Section. [18] Moreno, Ramon(1995): Macroeconomics Behavior during Periods of Speculative Pressure or Realignment: Evidence from Pacific Basin Countries, Economics Review, Federal Reserve Bank of San Francisco, N◦ 3, pp.3-16.
23
[20] Sachs, Jeffrey, Aaron Tornell and Velasco (1996) Financial Crises in Emerging Markets: The Lesson from 1995, Brookings Papers on Economic Activity, N◦ 1, pp. 147-215. [20] B. Bellone, E. Gautier and S. Le Coent (2005) Les march´es financiers anticipent-ils les retournements conjoncturels?,Notes d’´etudes et de recherche de la Banque de France, NER n◦ 128
24
