Economic Research and Perspectives

Economic Research and Perspectives

Early Warning Model and Currency Crisis in the Iranian Economy A Probit Approach

Authors
Mohammadjavad Khosrosereshki 1
Yavar Dashtbany 2
1 Ph.D. in Economics, Faculty of Management and Economics, University of Tarbiat Modares, Tehran, Iran
2 Ph.D. in Economics, Faculty of Economics, Islamic Azad University Science and Research Branch, Tehran, Iran
10.22034/24.4.161
Abstract
Aim and Introduction

In managed floating exchange rate systems, one of the important issues facing monetary policymakers is defining the exchange rate corridor and committing to it. In developing countries (with a low flow of foreign capital and limited access to international financial markets), it is difficult to follow committed policies. Iran is not only an oil-exporting country but also has limited access to export earnings due to sanctions. The central bank is heavily dependent on the government policies. Consequently, these causes have led to 5 currency crises since 1991. Therefore, having an early warning system for Iran’s exchange market is essential for monetary policymakers. One of the prominent aspects of this study is the definition of crisis index, which considers not only the direct intervention of the central bank in the exchange market but also both the exchange rate growth and the central bank’s foreign reserve growth rate. Finally, using the Probit approach, an early warning system with a low noise-to-signal ratio is proposed.

Methodology

Quarterly data from 1990 to 2023 for Iran and the United States were used in the model. Domestic data represent the real and monetary sectors of Iran’s economy, and U.S. data represent the world economy.

First, using Bai-Perron (1997) and oil revenue as a state variable of the economy, the duration was divided into three periods: low oil revenue, high oil revenue, and sanctions. The random walk equation was estimated.

logoil t =a+b logoil t-1 + ε t (1)

Second, for each period, according to Weymark's (1995) model, the exchange market pressure (EMP) and direct intervention indices were calculated. Then the average and standard deviation of both the exchange rate growth and the central bank’s foreign reserve growth rate were calculated.

Third, the situation in each quarter is determined as shown in Tables 1 and 2:
Table 1: The situation in each quarter in leaning with the wall intervention type







rt≥r-σr
rt<r-σr
Type of intervention:

leaning with the wall


Crisis type (1)
Mega-Crisis type (1)
et≥e+σe


Favorable
Ordinary
et<e+σe




Table 2. The situation in each quarter in leaning against the wall intervention type




rt≥r-σr
rt<r-σr
Type of intervention:

leaning against the wall


Ordinary
Mega-Crisis type (2)
et≥e+σe


Favorable
Crisis type (2)
et<e+σe



et is the exchange rate growth, and rt is the central bank’s foreign reserve growth. e and σe are the average and standard deviation of the exchange rate growth, respectively. r and σr are the average and standard deviation of the central bank’s foreign reserve growth, respectively. Ordinary and favorable situations are considered the same. But, the other situations are different.

Fifth, equation 2 was proposed as an early warning system.

 CRI t =α+ ß 1 CRI t-3 + ß 2 SCN t + ß 3 COV t + ß 4 FRG t-2 + ß 5 ERG t-2 + ß 6 ERGEN t-2 + ß 7 BFRLiq t-2 + ß 8 GBDLiqG t-2 + ß 9 GDBLiqG t-2 + ß 10 LiqG t-2 + ß 11 GDPG t-2 + ß 12 OilGDP t-2 + ß 13 OilG t-2 + ß 14 USAGDPG t-2 + ß 15 USAInf t-2 + ε t

All variables are I(0). CRIt is the quarterly situation. SCNt and COVt are the dummy variables for sanctions and COVID-19 respectively. The 2 lag variables are:

FRG is the central bank's foreign reserve growth. ERG is the exchange rate growth. ERGEN is the deviation of exchange rate growth from its long-term trend. BFRLiq is the ratio of the central bank's foreign reserve to liquidity. GBDliqG is the growth of the ratio of the government budget deficit to liquidity. GDBliqG is the growth of the ratio of the government debt to the central bank to liquidity. LiqG is the liquidity growth. GDPG is the growth of GDP. OilGDP is the ratio of oil revenue to GDP. OilG is the Brent crude oil growth. USAGDPG is the growth of the USA's GDP. USAInf is the USA's CPI.

Sixth, the forecasted variable is compared with the actual. Then, the NSR was calculated. NSR is between [0,1]. The less NSR, the better fitted the forecasted variable.

Findings

There were no crisis type 1 and also mega-crises type 1 and 2. Therefore, the situation in each quarter is either crisis type 2 or ordinary. The crises are 1994q1, 1995q2, 2012q1, 2012q4, 2018q3 and q4, 2020q1 to q4, and 2023q1.

The Probit model is: (2)

CRI t = -48.154 - 4.717* CRI t-3 + 1.718* SCN t + 11.496* COV t + 2.859* FRG t-2 + 302.002* ERG t-2 - 308.849* ERGEN t-2 - 40.659* BFRLiq t-2 - 36.540* GBDLiqG t-2 + 40.174* GDBLiqG t-2 + 99.098* LiqG t-2 - 14.066* GDPG t-2 + 98.677* OilGDP t-2 - 0.482* OilG t-2 - 4.932* USAGDPG t-2 + 9.820* USAInf t-2 + ε t

The above equation can predict all crises. The NSR is 0.143 which means that the model can good predict the crises.

Discussion and Conclusion

In this article, in addition to examining the exchange rate seasonal growth and the central bank's foreign reserves' seasonal growth, the central bank's performance in the face of currency crises has been determined. Then, a model to predict currency crises with the Probit approach is also provided. The results show that the central bank and other economic agents do not consider the same seasons as critical. The economic agents expected a crisis in any season in which the sanction is approved, regardless of the time of its implementation. In most of the currency crises, the central bank has aggravated the currency crisis by buying foreign currency to increase its foreign exchange reserves. The crisis period has not exceeded two seasons in all cases except the Corona period. Among the internal variables of the model, the variables of liquidity growth, exchange rate growth, foreign exchange reserve growth, oil income to GDP ratio, and sanctions are effective factors in predicting a currency crisis. GDP growth, the Brent crude oil growth, and the difference in exchange rate growth from the long-term trend are also more effective variables in reducing the possibility of currency crises. The proposed early warning currency crisis system could predict all currency crises. Finally, as a policy recommendation, it is suggested that instead of stabilizing the exchange rate level, the central bank should stabilize the exchange rate growth in the mid-term to increase its credibility among economic agents by reducing the inflationary environment and currency fluctuations
Keywords
Currency Crises
Early Warning System
Probit Model

Subjects

Al-Assaf، G. (2017). An Early Warning System for Currency
Crisis: A Comparative Study for the Case of Jordan and Egypt.
International Journal of Economics and Financial Issues 7(3): 43-50.
Ari، A. (2012). Early Warning Systems for Currency Crises: The Turkish Case. Economic Systems. 36: 391-410.
Ari، A. & Cergibozan، R. (2018). Currency Crises in Turkey: An
Empirical Assessment. Research in International Business and
Finance. https://doi.org/1016/10/j.ribaf.04/2018.001.
Babecký، J. Havránek، T. Matějů، J. Rusnák، M. Šmídová، K. & Vašíček، B. (2014). Banking، Debt، and Currency Crises in Developed Countries: Stylizedfacts and Early Warning Indicators. Journal of Financial Stability 15: 1-17.
Bai, Jushan, and Pierre Perron. (1998). Estimating and Testing Linear Models with Multiple Structural Changes. Econometrica 66: 47–78.
Bastanzad, H., & Davoudi, P. (2022). The Evaluation of Currency Crises in the Iranian Economy (Early Warning System Approach). Iranian Journal of Trade Studies, 26(103), 29-76. (in Persian) doi: 10.22034/ijts.2021.246928
Barzegar Marvasti, M. (2017). Leading Indicators of Currency Crises in Different Exchange Rate Regimes. Faculty of Economics, Management and Business. University of Tabriz
Berg، A. & Pattillo، C. (1999). Predicting Currency Crises: The Indicators Approach and an Alternative. Journal of International Money and Finance 18: 561-586.
Block، S. (2003). Political Conditions and Currency Crises in Emerging Markets. Emerging Markets Review، 4، 287-309.
Boonman، T. M. (2019). Dating Currency Crises in Emerging Market Economies. The North American Journal of Economics and Finance، 49، 273-286.
Castillo، F.A. (2006). Predicting a Currency Crisis Alternative Approaches and Applications to the Philippines، Master Thesis، Institutional Knowledge at Singapore Management University، Singapore Management University.
Chiodo، A.J. & Owyang، M.T. (2002). A Case Study of a. Currency Crisis: The Russian Default of 1998. Review، Federal Reserve Bank of St. Louis، Issue Nov، 7-18.
Ebrahimi, I. and Tavakolian, H. (2013). An Early Warning System of Currency Crisis in Iran: A TVTP Markov Switching Approach. The 22nd Annual Conference on Monetary and exchange Rate Policies Proceedings. Monetary and Banking Research Institute-Central Bank of Islamic Republic of Iran: 269-300. (in Persian)
Ferdous، Lutfa Tilat، Khnd Md Mostafa Kamal، Amirul Ahsan، Nhung Hong Thuy Hoang، and Munshi Samaduzzaman (2022). An Early Warning System for Currency Crises in Emerging Countries. Journal of Risk and Financial Management. https://doi.org/3390/10/jrfm15040167
Ghosh، S.، & Ghosh، A. (2002). Structural Vulnerabilities and Currency Crises. IMF Working Paper WP/09/02 ، Washington D.C.
Kaminsky، G.L. Lizondo، S. & Reinhart، C.M. (1998). The Leading Indicators of Currency Crises. IMF Staff Papers 45(1): 1-48.
Khosrosereshki M, Najarzadeh R, Heydari H. the Exchange Market Pressure and the Central Bank’s Direct Intervention in such a Market in Selected Oil-exporting Countries. qjerp 2022; 30 (101) :121-165
Nasiri N, Tayebi K, Shajari H, Vaez Barezani M. (2021). Modelliing of Earlly Warniing Indiicators of Currency Crisis: Emphasizing the Evaluation of the Relationship Between Currency Crisis and Capital Control Index. qjerp 2021; 28 (96) :91-120 (in Persian)
Nasrollahi, M., Yavari, K., Najarzadeh, R., & Mehregan, N. (2017). The design of an early warning system of currency crisis in Iran: A logistic regression approach. Journal of Economic Research (Tahghighat- E- Eghtesadi), 52(1), 187-214. (in Persian) doi: 10.22059/jte.2017.59617
Nasrollahi, M., Yavari, K., Najarzadeh, R., & Mehregan, N. (2021). Early Warning System for Currency Crises in Iran: The Markov-Switching Approach. The Journal of Economic Policy, 12(24), 107-138. (in Persian)
doi: 10.22034/epj.2021.10427.1831
Salmani, B., asgharpour, H., & KALAMI, M. (2019). Designing an Early Warning System for the Currency Crises in Iran Logit and Markov Switching Approaches. new economy and trade, 14(3), 97-124. (in Persian)
Weymark، D. (1995). Estimating exchange market pressure and the degree of exchange market intervention for Canada. Journal of International Economics, Vol. 39, 273–295.
Central Bank of Russia Annual Reports (1997-2003)
Central Bank of Russia Annual Reports (2014-2017)

Data Sources:
www.CBI.ir
www.fred.stlouisfed.org
Economic Research and Perspectives
Volume 24, Issue 4
Winter 2024
Pages 161-189