Stock Returns, Economic Growth, Interest Rates and the 2001 Crisis in Turkey

Transcription

1 The Empirical Economics Letters, 9(11): (November 2010) ISSN Stock Returns, Economic Growth, Interest Rates and the 2001 Crisis in Turkey Ülkem Başdaş and Uğur Soytaş * Middle East Technical University, Department. of Business Administration Ankara, Turkey Abstract: This paper investigates the relationship between economic growth and stock returns accounting for the interest rates and inflation in Turkey between 1997:1 and 2008:6. The Toda-Yamamoto procedure is used to test long run Granger causality between real stock returns, real activity, and interest rates. Empirical results show that over the period we study causality running from stock returns to real growth and from interest rates to real growth. After the 2001 crisis the link between real growth and real stock returns disappears completely. Hence, financial crisis seems to sever the link between stock market and the macroeconomy. Keywords: Stock Returns, Economic Growth, Crisis, Toda-Yamamoto Procedure JEL Classification Number: G10, O40, O53 1. Introduction The link between stock markets and macroeconomic variables has been widely studied. Early study by Goldsmith (1969) shows that there is a positive correlation between the financial intermediary assets as percentage of GDP and growth rates for 35 countries, including Turkey. Bosworth et al. (1975), Barro (1990), Fama (1981, 1990), and Schwert (1990) support the high correlation between stock returns and future real activity. An opponent approach suggesting that the liquidity of stock markets slow down the growth is explained by Demirgüç-Kunt and Levine (1996). Overall, there are no unanimous results on the link between macroeconomic variables and stock markets. As for Turkey, the number of studies is limited and results are inconsistent (Kargi and Terzi, 1997; Sari and Soytas, 2005; Karagoz and Armutlu, 2007; Erbaykal and Okuyan, 2007). The aim of this paper is to reexamine the link between real stock returns, real growth and real interest rates in Turkey over the period taking into account the 2001 crisis. This study tests the causality via Toda-Yamamoto procedure (Toda and Yamamoto, 1995), which has certain advantages over more traditional methods. First, it can be applied to series with any arbitrary integration order. Second, it does not require cointegration testing and therefore avoids carrying any bias in estimated cointegrating vector to further analysis. Third, since a VAR in levels is estimated there is no loss of information due to differencing. We show that after the 2001 crisis the causality between growth and stock returns vanishes. * Corresponding author. soytas@metu.edu.tr

2 The Empirical Economics Letters, 9(11): (November 2010) Literature Review An early study by Bosworth et al. (1975) demonstrates the leading role of stock returns for economic activity. Then, Fama (1981) finds significant links between stock returns, inflation, interest rates and real economic activity, confirming earlier results. He also reveals that the stock returns are positively related to real economic activity. In another study, Fama (1990) finds out that the growth rates of production explains 43 per cent of the variance in stock returns. Schwert (1990) strengthens the Fama s finding of positive relation between stock returns and real activity. James et al. (1985) shows that there is a strong positive link between stock returns and real economic activity. Barro (1990) confirms the significant explanatory power of changes in stock prices for investments in the U.S. and Canada. Empirical work does not always support a strong link between stock returns and growth. Canova and Nicolo (2000) consider the U.S., Japan, Germany and the U.K. markets from 1973 to 1995, but report significant results only for the U.S. Hassapis and Kalyvitis (2002) examine G-7 countries over and find out varying lengths of positive effects of real stock prices on growth. Mauro (2003) claims that the stock market should be taken into account to forecast output in both developed and developing countries. Binswanger (2004) examines the real stock prices, industrial production index and real GDP in several developed countries. The empirical findings suggest the link is broken recently not only for the U.S. but also for other selected countries. Enisan and Olufisayo (2009) focus on 7 sub-sahara African countries over the period Their results support the positive effect of market capitalization ratio and value traded ratio on per capita nominal GDP. They find bidirectional relation for Cote D Ivoire, Kenya, Morocco and Zimbabwe. Hassapis (2003) investigates the link among output, real stock price changes, interest rates, interest rate spreads and monetary aggregates for Canada and the U.S. for the period The results support the predictive power of stock returns for economic growth. Ritter (2005) considers the simple correlation between geometric mean annual real stock returns and arithmetic mean real per capita annual GDP growth for 19 countries over 33 years and 13 countries over 15 years. He makes a stronger claim stating that the correlation between real stock returns and per capita GDP growth is even negative over As for Turkey, there is also a lack of consensus (Kandır et al., 2007). Kargi and Terzi (1997) find out that industrial production index is effected only by its lagged values and nothing else for the period 1986:1-1996:6. Sari and Soytas (2005), on the other hand, find a positive relation between the real stock returns and real economic activity over 1986:1-2000:12. Karagoz and Armutlu (2007) utilize a bivariate model for the period 1988 to 2006 and show that economic growth Granger causes stock market growth. Erbaykal and

3 The Empirical Economics Letters, 9(11): (November 2010) 1061 Okuyan (2007) find Granger causality running from stock prices to economic growth and from interest rate to stock prices. We examine the role of 2001 crisis on the link between stock returns and growth in Turkey. 3. Data and Methodology This paper uses industrial production index, consumer price index, Istanbul Stock Exchange price index 100 (ISE 100) and monthly interest rates on deposits as proxies to real activity, inflation, stock returns and interest rates, respectively. Monthly industrial production index (IP) and monthly interest rate as weighted averages of 1-month deposits are acquired from the Central Bank of the Republic of Turkey based consumer price index is gathered from the Turkish Statistical Institute. Closing values of ISE 100 are sourced from Istanbul Stock Exchange. Stock returns, industrial production index and interest rates are in real terms. All series cover the period from 1997:1 to 2008:6. The TY procedure enables us to examine the relation among variables without testing for cointegration between the variables. Hence, a possible bias in the cointegration tests is not carried into the analysis. The basic requirement of TY procedure is to identify the maximum order of integration from the selected variables. To test the order of integrationfive different tests are adopted. Unit root tests show that RIR is integrated of order zero whereas RLIP and RLISE are integrated of order one (results are available upon request). When series are integrated of different orders Toda and Yamamoto (1995) procedure is more appropriate than the traditional methods. Assuming that the maximum order of integration is d, TY procedure requires choosing a VAR model with optimum lag length of k. Then, a lag augmented VAR(k+d) model is estimated. A modified Wald test on the joint significance of the first k coefficients of an independent variable is applied to test Granger causality. If the null of no joint significance is rejected, then that variable is said to have a long run causal impact on the dependent variable. The causality is interpreted as long run causality, since there are no first differences in the lag augmented VAR of the TY procedure. The VAR model for the system of real growth, real stock returns and real interest rates can be written as follows: p r s t α1 1i t i 1k t k 1l t l 1 1 1, t i= 1 k= 1 l= 1 p r s t α2 2i t i 2k t k 2l t l 2 2 2, t i= 1 k= 1 l= 1 p r s t α3 3i t i 3k t k 3l t l 3 3 3, t i= 1 k= 1 l= 1 (1) RLIP = + a RLIP + b RLISE + c RIR + d t +Ω DUM + e (2) RLISE = + a RLIP + b RLISE + c RIR + d t +Ω DUM + e (3) RIR = + a RLIP + b RLISE + c RIR + d t +Ω DUM + e

4 The Empirical Economics Letters, 9(11): (November 2010) 1062 where RLIP, RLISE, RIR denote real industrial production index, real stock prices and real interest rates, respectively. L represents the natural logarithm of a variable. represents the first difference of the corresponding variable. α's are the constant terms; e s are error terms; t is the trend term. Dummy variable DUM that takes 1 for 2001:2 and before; 0 otherwise. 4. Empirical Results As a first step of TY, an optimum lag length of 6 (k=6) is chosen according to AIC. Then the lag augmented VAR (6+1) model is estimated. The Wald test on first 6 coefficients of each variable in each equation is conducted for long run causal relationships. The Granger Causality test results are presented in Table 1. Over the period real interest rates and real stock returns Granger cause real growth, but not vice versa. Table 1: Granger Causality Test Results on VAR(7) (1997:1-2008:6) Equatıons Varıables Real Stock Market Index Real Rates b c Real Stock Market Index Real Interest Rates Interest Note: a, b, and c represent significance at the 1, 5, and 10% respectively. Significance implies that the column variable Granger causes the row variable. Diagnostic results are available upon request. This result partially supports the study of Kargi and Terzi (1997) but oppose the findings of Karagoz and Armutlu (2007). Different from the study of Erbaykal and Okuyan (2007), there is no Granger causality from interest rates to stock returns even though causality goes from stock market to growth. Our dataset considers a more recent time period that includes the 2001 crisis. None of the studies investigate the impact of economic or financial crisis on the relationship between stock returns and growth in Turkey. We additionally analyze the impact of the 2001 crisis. In that respect, we apply the TY procedure for the after crisis period starting from 2002:1 to 2008:6. Maximum order of integration is determined as 1 and VAR(5) is chosen based on AIC and diagnostic tests. Then, VAR(5+1) is run to investigate long run causality. The Wald Tests are carried out in Table 2.

5 The Empirical Economics Letters, 9(11): (November 2010) 1063 Table 2: Granger Causality Test Results on VAR(6) (2002:1-2008:6) Equatıons Real Stock Market Index Real Interest Rates Varıables Real Stock Market Index a c b - Real Interest Rates Note: a, b, and c represent significance at the 1, 5, and 10% respectively. Significance implies that the column variable Granger causes the row variable. Diagnostic results are available upon request. Comparison of Tables 1 and 2 underlines important differences. First, after the crisis the link between real stocks returns and real growth totally disappears. Second, both real growth and real stock returns lead the real interest rates after the crisis. Third, causality running from real interest rates to real growth persists. Full sample and post-crisis analysis suggests a change in the stock market-growth nexus in Turkey. The weak bi-directional causality over the whole period completely disappears after the 2001 Crisis. This result suggests that the 2001 crisis led to the detachment of real growth and real stock returns. According to the Capital Markets Board of Turkey Monthly Bulletin (January 2009), the foreign custody ratio increases well above 70% in 2007 from a low of 40% in The increasing foreign share enables us to understand the real interest rates equation. In place of real stock returns, the real interest rates, representing returns on alternative investments, may correlate with the real growth. This explains the finding of Granger causality from real stock returns to real interest rates. 5. Conclusions This paper adopts a recent dataset from 1997:1 to 2008:6 to investigate the causality among stock returns, real growth and interest rates in Turkey via the TY procedure. The results over the whole period show that there is weak bi-directional causality. On the other hand, the link between growth and stock returns disappears over 2002:1-2008:6 following the 2001 crisis. After the crisis causality from real stock returns and real growth to real interest rates becomes significant. It appears that financial crisis severs the link between stock market dynamics and the macroeconomy. One reason is the increased foreign share in the Istanbul Stock Exchange (ISE).

6 The Empirical Economics Letters, 9(11): (November 2010) 1064 References Barro, R.J., 1990, The Stock Market and Investment, Review of Financial Studies, 3(1), Binswanger, M., 2000, Stock market booms and real economic activity: Is this time different?, International Review of Economics and Finance, 9, Binswanger, M., 2004, Stock Returns and Real Activity in the G-7 Countries: Did the Relationship change during the 1980s?, Quarterly Review of Economics and Finance, 44, Bosworth, B., S. Hymans and F. Modigliani, 1975, The Stock Market and the Economy, Brookings Papers on Economic Activity, 2, Canova, F. and G.D. Nicolo, 2000, Stock Returns, Term Structure, Inflation, and Real Activity: An International Perspective, Macroeconomic Dynamics, 4, Capital Markets Board of Turkey, January 2009 Monthly Bulletin Enisan, A.A. and A.O. Olufisayo, 2009, Stock Market Development and Economic Growth: Evidence from Seven Sub-Sahara African Countries, Journal of Economics and Business, 61, Erbaykal, E. and H.A. Okuyan, 2007, Türkiye de Temel Makroekonomik Değişkenler ile Hisse Senedi Fiyatları arasındaki Nedensellik İlişkisi, İktisat, İşletme ve Finans, 22(260), Fama, E.F., 1981, Stock Returns, Real Activity, Inflation, and Money, American Economic Review, 71(4), Fama, E.F., 1990, Stock Returns, Expected Returns, and Real Activity, Journal of Finance, 45(4), Goldsmith, R.W., 1969, Financial structure and development, New Haven: Yale University Press. Hassapis, C., 2003, Financial Variables and Real Activity in Canada, Canadian Journal of Economics, 36(2), Hassapis, C. and S. Kalyvitis, 2002, Investigating the Links between Growth and Real Stock Price Changes with Empirical Evidence from the G-7 Economies, Quarterly Review of Economics and Finance, 42, James, C., S. Koreisha and M. Partch, 1985, A VARMA Analysis of the Causal Relations Among Stock Returns, Real Output, and Nominal Interest Rates, Journal of Finance, 40(5),

7 The Empirical Economics Letters, 9(11): (November 2010) 1065 Kandır, S.Y., Ö. İskenderoğlu and Y.B. Önal, 2007, Investigating the Relationship between Financial Development and Economic Growth, Çukurova University Social Sciences Institute Journal, 16(2), Kargi, N. and H. Terzi, 1997, Causal Relations among the ISE, Inflation, Interest Rates and Real Activity in Turkey: A VAR Analysis, ISE Review, 1, Karagoz, K. and R. Armutlu, 2007, Hisse Senedi Piyasasının Gelişimi ve Ekonomik Büyüme: Türkiye Örneği, 8. Türkiye Ekonometri ve İstatistik Kongresi, İnönü University, Malatya, Turkey. Mauro, P., 2003, Stock Returns and Output Growth in Emerging and Advanced Economies, Journal of Development Economics, 71, Ritter, J.R., 2005, Economic Growth and Equity Returns, Pacific-Bain Finance Journal, 13, Sari, R. and U. Soytaş, 2005, Inflation, Stock Returns, and Real Activity in Turkey, Empirical Economics Letters, 4(3), Schwert, G.W., 1990, Stock Returns and Real Activity: A Century of Evidence, Journal of Finance, 45(4), Toda, H.Y. and T. Yamamoto, 1995, Statistical Inference in Vector Autoregression with Possibly Integrated Processes, Journal of Econometrics, 66,