ارتباط پویا و تاثیرات نوسانات در بازار سهام بالکان

This paper investigates the presence of time-varying comovements, volatility implications and dynamic correlations in major Balkan and leading mature equity markets, in order to provide quantified responses to international asset allocation decisions. Since asset returns and correlation dynamics are critical inputs in asset pricing, portfolio management and risk hedging, emphasis is placed on the respective (constant and dynamic) equity market correlations produced by alternative multivariate GARCH forms, the Constant Conditional Correlation and the Asymmetric Dynamic Conditional Correlation models. The Balkan stock markets are seen to exhibit time-varying correlations as a peer group, although correlations with the mature markets remain relatively modest. In conjunction with sensitivity analysis on the asymmetric variance–covariance matrix, active portfolio diversification to the Balkan equity markets indicates to potentially improve investors’ risk-return trade-off.

A carefully managed enlargement policy has been at the heart of the European Union’s (EU) development over several decades, targeting economic and financial integration for Europe as a whole. The accession of Bulgaria and Romania on 1 January 2007 has completed the fifth enlargement, following the earlier accession of 10 Member States in May 2004. The new Members have introduced structural reforms to meet the institutional requirements of the EU guidelines and have managed to attract growing foreign direct investment flows (Bank of Austria, 2004). Future enlargement will concern a number of countries of South-Eastern Europe that are at various stages on their road towards the EU. Turkey and Croatia are candidate countries, as they have started accession negotiations on 3 October 2005. The new Members are expected to follow a stabilization and integration path into the EU. To this end, capital mobility and efficient equity markets play a critical role in the acceleration of the Balkan economic convergence process. For international investors, a key question remains whether fund allocation to the Balkan equity markets offers robust diversification benefits relative to developed equity markets and improves investors’ risk-return profile. In this line, the identification of the driving forces that shape stock returns, volatility behaviour and correlation dynamics turns to be a core issue for successful international asset allocation and efficient portfolio management. Cross-market linkages and spot-futures correlations are also particularly relevant to optimal dynamic hedging strategies and risk control. Past empirical literature has indicated a number of factors that may be responsible for emerging-mature market interrelationships, return comovements and volatility spillovers, including, indicatively, economic policy coordination, financial innovations and market deregulation, interest rate movements or financial crises with contagion effects (e.g. Syriopoulos, 2004 and Syriopoulos, 2006). Tight market linkages indicate that a domestic capital market may not be efficiently insulated from external shocks and international portfolio diversification benefits can be limited. A number of past financial studies deal with financial market linkages and shock contagion effects between mature and emerging equity markets (e.g. Aggarwal et al., 1999, Ratanapakorn and Sharma, 2002, Bessler and Yang, 2003, Chaudhuri and Wu, 2003, Wong and Vlaar, 2003 and Syriopoulos, 2004). Nevertheless, the Balkan equity markets have been neglected and empirical research remains thin on this region (Samitas et al., 2007). Some past studies indicate that emerging stock market behaviour increasingly depends on mature stock market volatility swings (e.g. Phylaktis and Ravazzolo, 2002, Swanson, 2003, Yang et al., 2003, Syriopoulos, 2006 and Syriopoulos, 2007). If this is the case, then the empirical issues at hand can have considerable adverse implications for efficient portfolio diversification and risk control. Given that correlations in international equity markets have increased over time (e.g. Chan et al., 1997, Capelle-Blancard and Raymond, 2002 and Marcelo et al., 2008), investors are interested in assessing the level of equity market linkages and the implications related to increased volatility and correlations, in order to design well-diversified portfolio strategies. A body of past literature attempts to quantify the implications of asymmetric dynamics on return volatility and correlation processes (e.g. Engle and Ng, 1993, Koutmos and Booth, 1995 and Bekaert and Wu, 2000), as asymmetric volatility spillover effects have been indicated to be present in major financial markets (e.g. Koutmos and Booth, 1995, Kroner and Ng, 1998, Yang and Doong, 2004 and Mazzotta, 2008). Similarly, the dynamic correlations between international capital markets also present asymmetric characteristics. Empirical evidence supports that correlations tend to increase in equity markets during turbulent periods or downward markets (e.g. Longin and Solnik, 2001, Ang and Chen, 2002 and Kearney and Poti, 2006). The core objective of this paper is to examine time-varying linkages and comovements between major Balkans equity markets and a number of leading developed equity markets; to identify key determinants of stock market co-risk and return within and across emerging and mature stock markets; to assess dynamic market correlation implications that are decisive inputs for international asset allocation, risk dispersion and efficient hedging; and, as a result, to build meaningful dynamic portfolio strategies. For this purpose, we initially implement and constructively compare the Constant Conditional Correlation (CCC) model (Bollerslev, 1990) and the Dynamic Conditional Correlation (DCC) model (Engle, 2002). Moreover, we estimate the Asymmetric Dynamic Conditional Correlation (ADCC) model, a structure that attempts to relax excessive parameter constraints of the earlier models, postulates a parsimonious parametric specification, nesting the CCC and DCC models, and allows for asymmetric effects in the variance functions (Cappiello et al., 2006). The proposed models are multivariate generalised autoregressive conditional heteroscedasticity (GARCH) structures that have gained empirical success recently in modelling time-varying asset correlations and large covariance matrices. To gain robust empirical insights, we carefully select our sample to include new EU members (Romania, Bulgaria and Cyprus), old EU members (Greece) and potential candidates (Turkey, Croatia) against benchmark mature markets (Germany and the USA). We focus our interest on the Balkan region for a number of reasons: for a start, contrary to other emerging markets, limited empirical research has been produced on the Balkan equity markets; a number of Balkan states have recently joined the EU or plan to become a member and this has important implications for international investors; the Balkan economies exhibit robust growth rates relative to mature counterparts, whereas the Balkan equity markets have shown impressive returns recently. At the same time, international private and institutional investors are in search of attractive style investment alternatives and are interested in niche asset classes and opportunities. Hence, this empirical exercise is rendered to be timely and useful. The paper is organized as follows. Section 2 presents data and statistics of the Balkan equity markets. Section 3 analyses the modelling methodology and Section 4 evaluates the empirical findings. Section 5 conducts a sensitivity analysis to build meaningful dynamic portfolio strategies. Finally, Section 6 concludes.

This study has undertaken an extensive investigation of the linkages, comovements and interdependences between major South European equity markets in the Balkan region, old, new or prospective EU members (Bulgaria, Romania, Croatia, Turkey, Cyprus, and Greece), and leading mature equity markets (US and Germany). Despite their rising importance in the international financial system, leading to growing foreign capital inflows, empirical research on the Balkan equity markets remains exceptionally thin. A core objective of this study has been to contribute a range of quantified responses to international portfolio investors that can potentially consider asset diversification to major Balkan equity markets and seek to improve their respective risk-return trade-off. The empirical results indicated the presence of one cointegrating vector in the sample equity markets, whereas the mature developed equity markets exert a significant long-term impact on the Balkan equity market behaviour. These findings point to potential comovements and volatility effects especially in turbulent market periods that may weaken potential long-run diversification benefits. Nevertheless, international investors can still find gainful opportunities in the Balkan equity markets over a short-run investment horizon. The variance–covariance analysis produced useful information on the dynamic correlations between Balkan and leading developed equity markets, on the basis of the Constant Conditional Correlation and, especially, of the Asymmetric Dynamic Conditional Correlation models. The empirical findings confirm, in most cases, the absence of any constant correlation between the equity markets under study. As anticipated, the developed equity markets were found to be highly correlated with each other. The Balkan equity markets, on the other hand, were seen to exhibit significant dynamic correlations as a peer group, whereas asymmetric correlations were also detected in some cases. Combined with the previous findings, these results may support attractive diversification opportunities for active international portfolio investors. Nevertheless, the asymmetric correlation increases, especially in case of negative news, may offer limited diversification benefits during downward markets. To gain a better insight into the Balkan equity market dynamics, we then proceeded to perform a sensitivity analysis on the estimated DCC variance–covariance matrix, comparing and contrasting it with alternative multivariate models, namely the OGARCH, EWMA and VSA models. The idea was to capture potential equity market interactions, as measured by the covariance matrix in each model and assess implications for risk control and portfolio performance. We constructed a universal portfolio with varying mature-emerging asset weights and concluded that the estimated conditional variances were gradually decreasing, whenever the Balkan equity markets were included in the universal portfolio. Dynamic portfolio diversification to the Balkan equity markets can then offer potential rewarding investment opportunities and improve investors’ risk-return profiles. Further empirical research on these issues would be useful, as the Balkan economies are catching up fast in their course towards European integration and economic growth prospects appear to be promising.