ISSN 1991-3087
Рейтинг@Mail.ru Rambler's Top100
Яндекс.Метрика

НА ГЛАВНУЮ

Разработка системы раннего предупреждения для оценки вероятности возникновения кризиса платежного баланса

 

Хомяк Василий Романович,

аспирант Киевского национального университета им. Тараса Шевченко.

 

Designing the early warning system for estimation the probability of the Balance of Payment crisis (BoP)

 

Khomiak Vasyl,

PhD student at the Taras Shevchenko National university of Kyiv.

 

История становления теорий относительно кризиса платежного баланса описана в статье. Проанализированы преимущества и недостатки применяемых подходов к моделированию системы раннего предупреждения кризиса. Описаны подходы формализации кризиса платежного баланса.

Ключевые слова: система раннего предупреждения, кризис платежного баланса, индекс валютного давления.

 

History of establishment theories of the balance of payment crisis is analyzed in the paper. Different methods and approaches of modeling the Balance of payment crisis are compared. Ways how to formalize crisis are mentioned

Keywords: Early warning system, Balance of Payments crisis, Exchange market pressure index.

 

Actuality

 

The first time the Balance of Payment crisis (BoP) mentioned at the beginning of the 70-th, than Paul Krugman presented his first paper dedicated to the problem of the BoP crisis in 1979. During the 80-th and 90-th due to deeper international integration and higher intention of financial flows actuality of modeling BoP crisis sharply increased. Few wages of BoP crisis pushed scientific community for a deeper research. In 1992 the French franc and the Irish punt came under attack as a result of the British pound and the Italian lira devaluation In 1994 BoP crisis started from Mexico caused the Latin America crisis. Unexpected Thailand BoP crisis in 1997 idled growth of “Asian tigers”, caused Asian BoP crisis that had significant contagion effect on partner countries. Russian BoP crisis in 1998 had significant effect on whole CIS countries. Global financial crisis in 1998 started in transition countries from the BoP crisis. The perfect example of hidden BoP crisis is recent situation in Eurozone. In fact, debt problems became a consequence of accumulation the current account deficit by southern European countries from the beginning of 2000. This accumulation compensated by financial inflow that stopped in 2008. It was a trigger of the BoP crisis.

The goal of this paper is analysis of instruments that can be used for design the Early Warning system that estimates probability of the BoP crisis. I hope I could prove to reader that problem of developing the Early warning system for predicting PoB crisis is no time. Because of fact that contagion effect of the BoP crisis becomes much stronger than earlier (the case of the East European countries) and the BoP crisis suits to be a trigger mechanism of financial and banking crisis.

 

Evolution of the BoP models

 

There are a plenty of literature about forecasting and preventing BoP crisis. The first theoretical model, proposed in 1979 (Krugman Paul ), is based on hypothesis that fixed exchange rate leads to the loss of international reserves and to speculative attack because of chronic deficit of the payment balance. Model suggests that the period preceding a currency crisis would be characterized by persistent decline in international reserves. According to proposed model, the evolution of the real exchange rate, the trade or current account balance, real wages and domestic interest rate could be a leading indicator of crisis. This type of model is named as the first generation model. The first generation model operates under the condition that a vicious cycle is created where decreasing foreign-exchange reserves can no longer finance a rising debt.

The second generation models were described precisely by Mariuce Obstfeld in 1994 and by Robert Flood and Nancy Marion in 1999. This type of models is based on hypothesis that effects of speculative attack or just decreasing in expectations can lead to crisis even without changes in fundamental indicators. Ozkan and Sutherland (1995) proposed models based on assumption that the list of factors that may affect the authorities is an objective function. This system of indicators could be used for alarming about currency crisis. Authors of model propose a system of indicators that in each specific case can indicate a higher likelihood of crisis. For example, Obstfeld (1996) used such indicators as domestic interest rate, public debt, central bank credit to banks, number of credits, political variables etc. Recent models argue that crises may develop without any noticeable change in economic fundamentals. According to this approach, the contingent nature of economic policy may generate self-fulfilling crises based on assumption that economic policies are not predetermined and crisis can occur only because of pessimistic expectations.

 

Table 1.

Approaches to designing the early warning system of currency crisis.

Authors

Year of publication

Countries in regression

Time range

Dependent variables

Independent variables

Fratzscher, Bussiere

2002

30 emerging countries

1989q1-1998q2

Exchange pressure

index

Capital flow, short-term capital flow, lending boom, foreign debt, short-term debt, overvaluation, reserves, trade balance, real contagion, bank contagion.

Kaminsky, Lizondo and Reinhart

1998

Noise-to signal approach, emerging countries: panel of 28 countries

Crisis episodes 1994/

1995, 1997/

1998

Crisis episodes

Real exchange rate, terms of trade, M2/reserves, lending rate/deposit rate, exports, bank deposits, international reserves, stock price index, excess M1, real interest differential, domestic, credit/GDP, current account/GDP, M2, imports, and industrial production.

Frankel, Jeffrey

2005

Latin America

1995-1999

Crisis episodes

Exchange rate, exporter prices, tariffs between countries, log wages, longterm inflation, long term; exchange rate volatility.

Reinhart

2001

Regressions are built to separate countries:

Spain, Thailand, Korea, Philippines, Malaysia

Changes in depends on country

Exchange pressure

Bid/ask spread, interest rate, change in interest rate, stock return, domestic interest rate changes, exchange rate.

Cuaresma, Slacík

2007

Crisis episodes 24 emerging Argentina, Brazil, Chile, Colombia, the Czech Republic, Ecuador, Egypt, Hungary, India, Indonesia, Israel, Korea, Malaysia, Mexico, Morocco, Peru, the Philippines, Poland, Russia, Slovakia, South Africa, Thailand, Turkey and Venezuela

January 1980 - December 2001

Binary variables, crisis episodes

Deviation of the real effective exchange rate from a Hodrick-Prescott trend, GDP growth rate, Short-term debt, current account balance relative to GDP, total debt relative to reserves, debt relative to reserves, Government balance, Primary fiscal balance relative to GDP, de facto freely falling FX regime, real interest rates, short term nominal interest rates deflated by the CPI, Broad money to reserves, annualized growth rate of real domestic credit, stock market, annualized growth of the composite stock market index.

 

The third type of model was outlined by Paul Krugman in 1999, focuses on the classic Keynesian idea of a “transfer problem”. And it is called “models of 3-rd generation”. This model is some kind of a synthesis of many of the conclusions drawn by the first and the second generation models are mentioned above. Krugman hypothesizes is based on the existence of three separate equilibriums: growth equilibrium, crisis equilibrium, and a transitional equilibrium. That synthesis based on models of earlier generations conducts a system of indicators.

All papers are mainly used two approaches. The first one is based on using econometrics models to check robustness of relations between variables and crisis. Next econometrics techniques are widely used in papers: probit, logit and OLS regression models. (Eichengreen and Rose). Another, alternative approach is proposed by Kaminsky and Reinhart in 1996, and Kaminsky, Lizondo and Reinhart in 1998 is known as the "signals" approach which essentially optimizes the signal to noise ratio for the various potential indicators of crisis. The idea of “signal approaches” is to calculate the number of cases, when variable crosses threshold. Crossing threshold of variable sends a “signal” about increasing likelihood of crisis..

There are a plenty of literature dedicated to the Early warning system, but this question is studied quite extensively. Despite of large number of researches dedicated to the problem of the early warning system, it is hard to receive useful lessons from empirical work. The definitions of a financial crisis and the severity of incidence vary widely. The literature investigates different types of crisis in different countries and over different time periods. The results from most of these studies therefore lack generality, and the lessons learned from one crisis and country may not be relevant for another. Also the empirical work on leading indicators faces a problem of selection bias.

Current problems in the Eurozone with the resent memory of 2008 global crisis caused a keen interest to the problem of developing early warning system and exploring contagion effect. Most of the new research papers are based on the fundamental papers that were written in 80-th and 90-th.

The first theoretical model proposed by Krugman ,1979 is based on hypothesis that fixed exchange rate leads to the loss of international reserves and to speculative attack because of chronic deficit of the payment balance. Model suggests that the period preceding a currency crisis would be characterized by a persistent decline in international reserves. According to the proposed model, evolution of the real exchange rate, trade or current account balance, real wages and domestic interest rate could be leading indicators of the crisis. This type of model is named as the first generation model.

The second generation models were described precisely by Obstfeld (1996), Flood and Marion (1999). This type of models stands on hypothesis that effects of speculative attack or decrease in expectations can lead to the crisis even without changes in fundamental indicators. Ozkan and Sutherland (1995) proposed models based on assumption that the list of factors that may affect the authorities is an objective function. This system of indicators could be used for alarming about the currency crisis. Recent models argue that the crises may develop without any noticeable changes in economic fundamentals. According to this approach, the contingent nature of economic policy may generate self-fulfilling crises based on assumption that economic policies are not predetermined and crisis can occur only because of pessimistic expectations.

The third type of model outlined by Paul Krugman in 1999 is focused on the classic Keynesian idea of a “transfer problem”. It is called “model of 3-rd generation”. This model is some kind of a synthesis of many conclusions drawn by the first and the second generation models mentioned above. The era of designing early warning system began in 1990s when a wide range of methodological debates started, including the balance-of-payments crisis (Kaminsky and Reinhart, 1996).

A binary choice models, such as probit/ logit, were widely used. Examples of these studies are Eichengreen et al. (1995, 1996) and Frankel and Rose (1996). Recently, Berg and Pattillo (1999), Komulainen and Lukkarila (2003) and Kumar et al. (2003) have also analyzed the predictability of emerging markets currency crises using probit/logit models, whereas Bussière and Fratzscher (2002) used a more sophisticated multinomial logit model. The certain economic fundamentals were examined in studies that can explain currency crises and the crises of the 1980s and the 1990s would have been, at least to some extent, predictable. Another method is to consider the predictive power of the variables one at a time (univariate) so that a variable is considered as a good leading indicator giving a correct signal of crisis before the incident.

The early warning system of 1990s (Kaminsky, 1999) is concerned more about developing countries, while later papers of Rose and Spiegel, 2009 included large sample including developed countries as well.

Another issue discussed in papers was prediction power of the early-warning system. Too high noise-to-signal ratios to predict the future crises credibility in the eyes of policy makers (Berg and Pattillo, 1998) is characterized in many studies. A wide range of new tools was applied to solve this problem using new techniques such as Markov-switching (Peria, 2002 and Abiad, 2003) or multinomial logit models (Bussiere and Fratzscher, 2006). In addition, it offers policy makers an explicit choice to preselect their preferences regarding the missed crisis and false alarms (Alessi and Detken, 2009).

Most of the latest researches concentrated on analyzing crisis 2008. Obstfield, Shambaugh and Taylor (2009, 2010) were among the first to investigate incidence of the crisis. Percentage change of the exchange rate of national currency against the US currency and found that excess of reserves deviated by M2 that was a significant predictor. More wide research was made by Rose and Spiegel (2009a, 2009b) who model crisis as changes in real GDP, stock market, country credit rating and the exchange rate. They show high significance of contagion effect

 

Designing the Eraly warning sytem

 

Contagion effect is analyzed as a leading indicator of the crisis by (Frankel and Saravelos, 2010) and (Bussiere and Fratzscher, 2006). Some measures allow to infer systemic risk contribution as well. Similarly, Hartmann, Straetmans, and de Vries (2005) derive indicators of the severity of banking system risk from banks’ equity returns using multivariate extreme value theory.

One of the key questions in crisis modeling is selection of dependent variable. Using exchange market pressure (EMP) index deals with formalizing crisis. There are few approaches how to calculate this index, majority of them includes two major components: exchange rate and international reserves. This approach (Kaminsky, Lizondo, Reinhart (1998)) is more widely used for modeling the early warning system:

                                                                                 (1)

where rm – reserves of the National bank of Ukraine in international currency; - standard deviation of resaves of the National bank of Ukraine; ei,t – Real effective exchange rate; - standard deviation of REER.

The first one, proposed by Kaminsky (1996) is noise-to signal approach. This approach is based on the variety of monthly indicators that signal crisis if one of them crosses threshold. The information from each variable is complexly analyzed and evaluates probability of a forthcoming crisis, each indicator has 4 states, Table 2.

 

Table 2.

Components of indicators in noise-to-signal approach.

 

Crisis held during t period

Crisis doesn’t held during t period

Crisis is signaled

A

B

Crisis is not signaled

C

D

 

To calculate goodness-of-fit for indicator the noise-to-signal ratio (KLR ) statististic should be compared. Empirically it is proven, that if KLR statistic is less than 1, indicator can be used in the EWS system, if it is more than 1 – it is recommended to eliminate it.

                                                                                               (2)

Econometrics techniques are used to select the set of indicators to include to the early warning system as well. Exchange Market Index is used as dependent variable. Regression model was proposed by Frankel and Rose (1998). Model indicates a broad set of potential indicators. It is based on hypothesis that sharply changing in capital inflow causes to currency crashes. Estimation of the model and checking hypothesis shows that high credit growth, low share of reserves, overvaluated real effective exchange rate, high fiscal and current account deficits, high foreign interest rates are significant indicators of the crisis.

There are some another approaches to compose exchange market price index. The main idea is to include three main variables: exchange rate, reserves and interest rate. The combination of these three variables composes the best indicator that helps to formalize measuring of economic performance.

Another technique is to build a discrete variable. Discrete variable equals one in the period when the exchange market index crosses the threshold. Threshold can be given as a nominal level, but more commonly for threshold is to take changing during the period. For example, if exchange rate market index changes per quarter for 5%, it means that discrete variable becomes equal to 1. If exchange rate market index does not cross threshold, it equals zero. Then logit or probit models can be applied to estimate probability of crisis and main factors that lead to it.

Another technique is Markow's approach (Abiad 2003) that is efficient to model of rational expectations or contagion from another countries. The main reason to use Markow approach is because it gives opportunity to be more flexibly about assumptions that are required in standard models. Small devaluation in speculative index can foreshadow upcoming crisis. The variable in the model follows a first-order, two-state Markov chain , where st=1denotes a crisis state, st=0 denotes a tranquil period. S is determined by the nominal exchange rate or by the speculative pressure index. This variable is normally distributed. The density of yt condition on st is then

exp() for st =0,1                                                (3)

The latent regime-switching variable st evolves according to the transition probability matrics Pt

 

Time t-1

Time t

 

State 0

State 1

State 0

State 1

 

where pij - is the probability of going from state i in period t-1 to state j in period t , and F is cumulative distribution function. Matrices pij should be estimated by the help of experts or on assumption how contagion effect in country i influences country j .

            To summarize, modeling of the BoP crisis should include the next groups of independent variables: changes in fundamentals, contagion effect and rational expectations.

   (4)

where y is a dependent variable that describes crisis vi(si) – shows rational expectations among market players xt - fundamental changes in economics.

REALij and FINij shows contagion effect to BoP crisis. REALij shows how BoP crisis from country j could infect country i with trade links. FINij shows how BoP crisis from country k can infect country i with financial links.

Before regression analysis and estimating probability of crisis by logit model, facror analysis (Jakob, Lestano 2003) will be run. After that we will have few principal components that represent main groups of macroeconomic fundamentals (as in research (Jakob, Lestano 2003) can be received components that represent external, financial, internal and global factors). Than principal component are used in logit regression as independent variable that solve multicollinearity and endogeneity problem.

Special attention I would like to concentrate about modeling contagion effect of spreading crisis. If analyze last crisis, we can observe crisis contagion effect starting since 1992 when the French franc and the Irish punt came under attack as a result of the British pound and the Italian lira devaluation. Contagion effect in 1994 had more impact starting from Mexico crises infected Argentina without no signals about changing in fundamentals. During the next crisis of 1997 contagion effect was much more powerful. Starting unexpectedly from Thailand it hit whole region during 2 months. This examples show that due to deep economic integration the BoP crises could came from abroad without no changes in fundamentals. There are 2 channels how contagion effect can be transmitted: trade channel and financial channel. Trade link means that countries which trade disproportionately with one another are prone to contagion operating through the competitiveness effects of the crisis-induced exchange rate changes or decrease in demand on imported production. Financial link works in such way: crises in the country with similar risk averse lead investors to revise their expectations. To model trade link of contagion effect gravity models will be used linearized form (Fidmuc, Karaja, Tichit 2012).

(5)

YitBoP crisis in country i ; Yj binary variable that shows if BoP crisis happened in partner trading country j; Yk – binary variable that shows if BoP crisis happened in countries that are main investors of Ukrainian economy k, I would like to pay your attention that j should no be equal to k, but it it can happen sometimes; Distanceij – distance between country i and j; Contiguityij – dummy variable controlling for the presence of a common border between the two countries; Tradingpartner – share of export from country i to contry j, based on last year.

 

Conclusion

 

BoP crisis could have more negative impact than earlier because of deeper integration between countries and increase financial flows between countries. In fact, as can be noticed from crisis in 90-th, BoP crisis is a trigger of financial and banking crisis. The last could initiate the global crisis.

Current research shows that Early warning system succeeded with prediction the BoP crisis. Outlook of researches shows that changings in macroeconomic fundamentals indicate about increasing probability of crisis. Especially such variables as ratio of current account to gdp, ratio of financial account to gdp, monetization level and changing in interest rate successfully inform about crisis.

The tendency of last years also signals about increase of contagion effect impact. It means that even if no macroeconomic fundamentals signal about crisis, it could be imported from trading partner countries. Problem of modeling contagion effect is highly actual and will be researched in next papers.

 

Literature

 

1.                  Bussière M., 2007. "Balance of payment crises in emerging markets - how early were the “early” warning signals?" Working Paper Series 713, European Central Bank.

2.                  Cocozza, E. and Piselli, P. “Testing for east-west contagion in the European banking sector during the financial crisis”, Bank D’ITALIA 2011.

3.                  Cuaresma C, Jesýs & Slacik, Tomas, 2007. "An "almost-too-late" warning mechanism for currency crises," BOFIT Discussion Papers 4/2007, Bank of Finland, Institute for Economies in Transition.

4.                  Eichengreen B. & Rose, A 1998 "Staying Afloat When the Wind Shifts: External Factors and Emerging-Market Banking Crises," NBER Working Papers 6370, National Bureau of Economic Research, Inc.

5.                  Eichengreen, Barry & Rose, Andrew K & Wyplosz, Charles, 1996. “Contagious Currency crisis”, CEPR Discussion Papers 1453, C.E.P.R. Discussion Papers.

6.                  Fidrmuc J., Karaja E. and Tichit A., Reform, Uncertainty and Spillovers a Gravity Model Approach February 2012.

7.                  Flood, Robert & Marion, Nancy, 1999. "Perspectives on the Recent Currency Crisis Literature," International Journal of Finance & Economics, John Wiley & Sons, Ltd., vol. 4(1), pages 1-26, January.

8.                  Frankel J. and Sarvelos G, Are leading indicators of financial crises useful for assessing country vulnerability? Evidence from the 2008-09 global crisis, National bureau of economic research.

9.                  Frankel, Jeffrey, 2005. "Contractionary Currency Crashes In Developing Countries," Working Paper Series rwp05-017, Harvard University, John F. Kennedy School of Government.

10.               Graciela L. Kaminsky & Carmen M. Reinhart, 1996. "The twin crises: the causes of banking and balance-of-payments problems," International Finance Discussion Papers 544, Board of Governors of the Federal Reserve System (U.S.).

11.               Jeffrey A. Frankel & George Saravelos, 2010. "Are Leading Indicators of Financial Crises Useful for Assessing Country Vulnerability? Evidence from the 2008-09 Global Crisis," NBER Working Papers 16047, National Bureau of Economic Research, Inc.

12.               Krugman P., Kenneth S. Rogoff, Stanley Fischer and William J. McDonough, 1999. "Currency Crises" , NBER Chapters, in: International Capital Flows, pages 421-466 National Bureau of Economic Research, Inc.

13.               Krugman, Paul, 1979. "A Model of Balance-of-Payments Crises," Journal of Money, Credit and Banking, Blackwell Publishing, vol. 11(3), pages 311-25, August.

14.               Lucia Alessi & Carsten Detken, 2009. "Real Time’ early warning indicators for costly asset price boom/bust cycles - a role for global liquidity," Working Paper Series 1039, European Central Bank.

15.               Obstfeld M., 1994. "The Logic of Currency Crises," NBER Working Papers 4640, National Bureau of Economic Research, Inc.

16.               Rose A. & Spiegel M, 2009. "Cross-Country Causes and Consequences of the 2008 Crisis: Early Warning," NBER Working Papers 15357, National Bureau of Economic Research, Inc.

 

Поступила в редакцию 25.01.2013 г.

2006-2019 © Журнал научных публикаций аспирантов и докторантов.
Все материалы, размещенные на данном сайте, охраняются авторским правом. При использовании материалов сайта активная ссылка на первоисточник обязательна.