International interdependence of an emerging market: the case of Iran

Abstract

In this study the interdependence between Iran, its major trading partners and the United States is investigated using vector autoregression, generalized impulse response function and generalized variance decomposition techniques, introduced by Pesaran and Shin (1998). These techniques have an advantage over the commonly used impulse response and variance decomposition procedures in that they are insensitive to the ordering of the countries considered and hence, they produce more reliable results. The countries included in the sample, besides Iran are, France, Germany, Spain, Japan, South Korea, Brazil, Italy and the United States. The direction, strength, durability and stability of the effect of shocks in one market on the return patterns of the other markets are examined. The findings are 4-fold. First, the effect of past own market shocks on current behaviour is significant, beyond the first month, in most cases. Second, the own effect is stronger for the emerging markets such as Iran and Brazil, than the industrialized countries. Third, cross-country effects are short-lived for Brazil, Korea and Japan, but durable in the case of Iran, Germany, Spain and the United States. Fourth, in terms of breadth and strength, cross-country effects exhibit differential degrees of interdependence and asymmetry. The observed lack of integration between the Iranian market and the industrialized world makes it less vulnerable to the effect of shocks in the latter countries but it also deprives it from the flow of funds that could spur economic development and growth.

Acknowledgements

Thanks are due to Nader Habibi, Iqbal Mansur and an anonymous referee of the Journal for helpful comments and suggestions. Any remaining errors are ours. An earlier version of this article was presented at the annual meeting of the Center for Iranian Research and Analysis (CIRA) 2005, in Philadelphia, PA.

Notes

¹ Detailed description of the TSE is available from the authors. Related matters are also discussed in different issues of ‘Iran International’, 2004 and earlier. The Iranian market is considered to be an emerging market because it is relatively small in size and the Iranian economy is still in the stage of development. Less than 1% of the Iranians actively participate in the stock exchange while the figure is close to 60% in Europe (Payvand's Iran News, 12/22/03). Baier et al . (2003) show that opening a stock exchange increases economic growth. However, they find Iran to be an outlier in this regard, possibly because the Iranian revolution, which followed the opening of the stock exchange, led to a major decline in output. They argue that the revolution may be interpreted as a consequence of the opening of the stock exchange.

² It is notable that few studies have used the GVAR, GIRF and GVD techniques to examine the issue of interdependence across any set of markets. The few studies using them have not used stock market data. The sensitivity of the results to the variable ordering is briefly discussed is the methodology section.

³ The results of many studies based on the traditional IRF and VD change substantially in response to changes in ordering of the countries and are, therefore, suspect.

⁴ Kodres and Pritsker (2002) provide a good review of this literature.

⁵ The markets examined are Australia, Canada, France, Germany, Hong Kong, Japan, Switzerland, the United Kingdom and the United States.

⁶ A number of studies focus on the effect of crises and turbulence on the interdependence across markets. Some examples are Malliaris and Urrutia (1992), Arshanapalli and Doukas (1993), Arshanapalli et al . (1995), Hassan and Naka (1996), Choudhry (1996) and Masih and Masih (1997). The findings of these studies generally denote an increase in interdependence under crisis condition, and significance of interdependence in the post-crisis era. For Iran, it would be interesting to examine the effect of the Iranian–Iraqi war on market interdependence. However, the Iranian market was closed during the war and even for the early post-war period. Hence, the data necessary for this purpose are not available.

⁷ Elyasiani and Mansur (2003) examine spillover across the banking markets of the United States, Japan and Germany. The finding is that the spillover effects of volatility in interest rate between the United States and Japan (Germany) are asymmetric, with the United States manifesting itself as a leader. Moreover, unsystematic shocks that harm the banking sector of the country they originate from, are beneficial to the others, namely that banks of different countries act as competitors.

⁸ The literature on the TSE is seriously limited and published outside the mainstream journals. A search of the literature using the common search engines produces only one article, Sadeghi (1997). Sadeghi investigates whether the Iranian stock market serves as a hedge against inflation within a regression model, using monthly data over the 1991–1995 period. The findings based on regression analysis indicate that the Iranian stock market does not serve as a good hedge for inflation. Results are also drawn using t-tests for significance of mean monthly, annual and 5-year real returns on gold, foreign currency and equity shares. None of the tests is found to be significant, indicating that these investment vehicles produced zero real returns during the sample period (Table 2). Even the cumulative returns on these investment vehicles are most often insignificantly different from zero. Specifically, the cumulative real returns on gold and foreign exchange never become significant and the cumulative real returns on shares is insignificant, except for a brief period, May 1991 to December 1992 (Table 4). There are also some studies of the Iranian currency and money markets, e.g. Bahmani-Oskooee (1996a, b, 2002) and Bahmani-Oskooee and Shiva (1998).

⁹ See Hurley and Santos (2001) for an example of application of such orthogonalization.

¹⁰ See Bernanke (1986), Blanchard and Watson (1986), Sims (1986), Shapiro and Watson (1988) and Blanchard and Quah (1989).

¹¹ Since the details of traditional VAR, GIRF and GVD analysis are available in the literature, they are not repeated here. See Koop et al . (1996), Pesaran and Shin (1998) and Potter (2000) for full description of these techniques. See Dekker et al . (2001), and Elyasiani et al . (2004) for applications of these techniques to financial markets. To select the number of lags, several lag selection criteria including the sequential modified likelihood ratio (LR) test statistic, final prediction error (FPE), Akaike information criterion (AIC), Schwarz information criterion (SC), and Hannan–Quinn (HQ) information criterion are used. The results of these procedures are found to be inconsistent. Since none of the procedures is clearly superior to the rest, it is desirable to err on the side of a larger, rather than a smaller, number of lags because omitted variable problems due to exclusion of relevant lags would be much more serious than inclusion of possibly redundant ones (Kmenta, 1986). Following this principle, we choose three monthly lags for each country in the model.

¹² TSE produces an aggregate index called the All Share Index (TEPIX). This index is not considered to be a true representative of the current state of the market due to changes in the mix of the relevant and highly traded stocks. The IFS indices generally relate to common shares of companies traded on national or foreign stock exchanges. Monthly indices are obtained as simple arithmetic averages of the daily or weekly indices, although in some cases mid-month or end-of-month quotations are used. All reported indices are adjusted for changes in quoted nominal capital of companies. Indices are in general base-weighted arithmetic averages with market value of outstanding shares as weights.

¹³ We also obtained daily data on individual firms from the TSE to investigate the interdependence across the markets in the sample. Several problems led us to abandon this effort. First, and most importantly, many listed firms exhibited highly infrequent trading. Indeed, stocks of a large number of firms were not traded for months. As a result, dependable daily data could not be constructed for firms, and even portfolios of firms. Second, daily data for the TSE stocks are subject to a considerable degree of noise. This produces unreliable estimates of measures of market interdependence. Third, the speed of reactions in emerging markets such as the TSE to the world events is small, making lower frequency data suitable for analysis.

¹⁴ See Phillips and Perron (1988).

¹⁵ Elliot et al . (1996), and subsequent studies have shown that the DF-GLS test has a significantly higher power than the previous versions of the ADF test. The PP test performs a regression of the variable on its lags. The lag number is determined by the procedure as 4 _* (N/100)^(2/9), where N is the number of observations in the series. This test tends to be robust to a wide range of serial correlation patterns and heteroskedasticity.

¹⁶ The test-statistics for the test results are not reported here to save space. Country return autocorrelation, partial autocorrelation, and cross-market correlations are also calculated (not reported). According to these figures, autocorrelations for France, Germany and Spain are not significant at any lag, they are significant at 5% level for a few months in the case of Japan, Italy and the United States, and they persist up to 2 years for Iran, Brazil and Korea. Pair-wise correlations are small and insignificant in many cases.

¹⁷ Own market forces include e.g. shocks to the underlying economy, domestic speculation about the stock market and government interest rate and exchange rate policy decisions. Given that only one value is generated for VD shares, a formal test of hypothesis of segregation against the alternative of interdependence is not possible in the VD framework.

¹⁸ In a similar vein, Klaassen (2000) has found that, since the 1980s, currency markets in the western world have moved toward increased segregation (looser links), rather than further integration, due to increased volatility in these markets.

¹⁹ As a recent example of religious forces driving political and economic actions on the part of Iran, the Iranian president, Ahmadinejad, called for Israel to be wiped out from the face of the earth (CNN.Com, 27 October 2005). The president then added: ‘And God willing, with the force of God behind it, we shall soon experience a world without the United States and Zionism,’ according to a quote published by IRNA. One million Iranians marched in support of Ahmadenejad's position the same week. Iran has also barred trade with Israel since the Iranian revolution of 1979.

²⁰ This loss in information has to be balanced against the noisiness of the data in higher frequency sampling.

²¹ GIRF table is too long and not included here.

²² The values for other lags are left out to make the table more readable.