سرمایه گذاری در بازارهای سهام اروپا برای شرکت های فن آوری با پیشرفته

We used a recursive modeling approach to study whether investors, in real time could, have used information on the comovement of stock markets to forecast stock returns in European stock markets for high-technology firms. We analyzed weekly data on returns in the Neuer Markt, the Nouveau Marché, the Alternative Investment Market, and the NASDAQ. We found substantial changes over time in the usefulness of the inter-European and cross-Atlantic comovement of stock markets for predicting stock returns. We also studied how monitoring the comovement of stock markets would have affected the performance of simple trading rules and the investors' market-timing ability of investors.

At the end of the 1990s, advances made in high-technology sectors like the IT sector and the bio-sciences sector were in the focus of the mass media and investors. Investors were strongly interested in investing in high-technology firms that needed capital to finance their expansion. As a result, European stock exchanges founded new stock markets for high-technology firms. In Frankfurt, Paris, and London important marketplaces for trading stocks in European high-technology firms were established. A key problem of investors who planned to invest in the new European stock markets for high-technology firms was that little was known about these markets and the firms listed in these markets. Because these markets were new, investors knew little about how these markets processed information and how they reacted to news. Moreover, because many high-technology firms operated in completely new technological fields, investors had hardly any experience in assessing the growth prospects for firms listed on European stock markets for high-technology firms. As a result, investors' beliefs concerning the bright growth prospects of particular high-technology firms resulted in the bubble-like phenomena that were a characteristic feature of stock markets for high-technology firms in the late 1990s. In general, there was no empirical evidence available that could have helped investors to determine the key driving factors of stock returns in the new European stock markets for high-technology firms. Even worse, the potential for portfolio diversification across markets was limited because European stock markets for high-technology firms witnessed a non-negligible degree of comovement at the end of the 1990s. This comovement may even indicate that the portfolios held by investors who invested in these markets were vulnerable to the kind of contagion effects and spillovers of market jitters that have been widely studied in the recent literature (Forbes and Rigobon, 2002 and Hon et al., 2007). Contagion and spillover effects, however, do not necessarily imply that the comovement of stock markets was per se bad for investors. In fact, even investors who only invested in their domestic stock market for high-technology firms, rather than in international stock markets, may have benefited from the comovement of stock markets. The reason for this is that comovement of stock markets need not reflect only contemporaneous links between stock markets. Rather, comovement could also indicate that potentially complex lead-lag links between stock markets exist. If this is the case, comovement of stock markets could imply that investors can use international stock returns to predict returns in their domestic stock market. If comovement implies predictability of returns, this may even help investors to set up profitable simple trading rules based on the comovement of stock markets. While many authors have empirically studied the degree and the sources of the international comovement of stock markets (Bekaert and Harvey, 1995, Chinn and Forbes, 2004 and Longin and Solnik, 1995, among others), empirical evidence is relatively silent with respect to the comovement of stock markets for high-technology firms. Even less is known about the question whether investors who invested in these stock markets could have taken advantage of the comovement of stock markets for high-technology firms in order to increase the performance of their stock market investments. To the best of our knowledge, our study is the first empirical study to address the question whether investors could have used the comovement of European stock markets for high-technology firms to increase the performance of their investments. In order to conduct our empirical study, we used the recursive modeling approach developed by Pesaran and Timmermann, 1995 and Pesaran and Timmermann, 2000. A recursive modeling approach implies that, to predict stock returns, investors can only use a set of information that is available in the period of time in which investors have to reach investment decisions. Included in this set of information is the information on the international comovement of stock markets available in the period of time when investment decisions had to be reached. Not included is information on the comovement of stock markets in later periods of time. Thus, a recursive modeling approach renders it possible to explicitly account for the uncertainty concerning the comovement of stock markets that is a crucial aspect of investors' decision problem in real time. A recursive modeling approach has two further key advantages. First, a recursive modeling approach renders it possible to trace out potential changes in the comovement of stock markets over time. We deem this to be an important advantage because Hon et al. (2007) have recently reported empirical evidence of structural breaks in the comovement of stock index returns in the information technology and telecommunications sectors. In order to account for structural breaks, we split our dataset into a pre-crash subsample, which covers the time during the stock market bubble, and a post-crash subsample. Second, a recursive modeling approach renders it possible to analyze whether the comovement of stock markets could have been used by investors for the purpose of out-of-sample forecasting of stock returns in real time. A detailed analysis of such forecasting informs about whether investors could have exploited stock return predictability to set up profitable simple trading rules. Thus, our study adds to the recent studies of out-of-sample predictability of stock returns (see inter alia, Breen et al., 1990, Pesaran and Timmermann, 1995, Pesaran and Timmermann, 2000, Bossaerts and Hillion, 1999, Goyal and Welch, 2003, Fong and Yong, 2005 and Cooper et al., 2005). In order to study whether investors, in real time, could have exploited the comovement of European stock markets for high-technology firms, we compiled data for three European stock markets for high-technology firms: the Neuer Markt in Germany (founded in 1997), the Nouveau Marché in France (founded in 1996), and the Alternative Investment Market (AIM) in the United Kingdom (founded in 1995). It is interesting to study these markets for at least three reasons. First, empirical evidence regarding the implications of their comovement for the predictability of stock returns in real time is not available. Second, the recent literature on the comovement of stock markets has focused mainly on countrywide stock market indexes (Ehrmann et al., 2005) that are dominated by large and internationally active firms. The comovement of stock markets for high-technology firms, which are often small and domestically operating firms, might be very different from the comovement of stock markets for large and mature firms. Third, the prices of the stocks listed on European stock markets for high-technology firms rallied and crashed in the late 1990s. This led to substantial reorganizations of these markets over time. It should be interesting to analyze whether these reorganizations have had an impact on the comovement of stock markets. Our estimation results show that the comovement across European stock markets for high-technology firms significantly varied across stock markets and over time. Moreover, the relative usefulness of inter-European comovements as compared to cross-Atlantic comovement with the NASDAQ, the leading U.S. market for high-technology firms, also varied across stock markets and over time. For example, in the pre-crash subsample, we found evidence of comovement of the NASDAQ and the Neuer Markt and the Nouveau Marché, but not of the NASDAQ and the AIM. In the post-crash subsample, by contrast, the comovement of the NASDAQ and the Neuer Markt and the Nouveau Marché lost usefulness, while the comovement of the NASDAQ and the AIM gained in usefulness. Interestingly, we found that only in a few cases would investors have been able to use the comovement of stock markets for high-technology firms to increase the real time performance of simple trading rules. Thus, investors could not systematically exploit the comovement of stock markets for high-technology firms to set up trading rules that outperform trading rules that do not account for the comovement of stock markets. Finally, we found that taking information on the comovement of stock markets into consideration when forecasting stock returns does not systematically affect the market-timing ability of investors. We structure the remainder of our paper as follows. In Section 2, we describe the recursive modeling approach that we used to model how investors may have predicted stock returns in high-technology firms in real time. In Section 3, we describe the dataset we used in our empirical analysis. In Section 4, we report our empirical results. In Section 5, we provide some concluding remarks.

We analyzed the predictability of returns in European stock markets for high-technology firms. Using a recursive modeling approach, we documented how the usefulness of accounting for the comovement of stock markets for predicting stock returns changed over time. Our results indicate that the optimal forecasting models include NASDAQ returns for the Neuer Markt and the Nouveau Marché more often in the pre-crash period than in the post-crash period. In the post-crash period, the AIM returns became more useful than NASDAQ returns for forecasting purposes. In contrast, the comovement of NASDAQ returns and the returns in the AIM were more pronounced in the post-crash period than in the pre-crash period. We also analyzed, in terms of terminal financial wealth and in terms of Sharpe's ratio, the implications of changes in the comovement of stock markets for the performance of simple trading rules. These implications varied substantially across model-selection criteria and subsamples. It is, thus, not possible to give a universally applicable, simple answer to the question whether investors should account for information on the comovement of stock markets for high-technology firms when reaching their investment decisions. Finally, we found that accounting for information on the comovement of stock markets in general was not useful for market-timing purposes. How can our results be interpreted in economic terms? Our results suggest that the answer to this question differs across the pre-crash and the post-crash subsample. Our result that strong cross-Atlantic return links existed between stock markets for high-technology firms in the pre-crash period for the Neuer Markt and the Nouveau Marché, but not for the AIM, could be due to the industry composition of the European indexes and the over-evaluation of IT stocks at the end of the 1990s. Such an interpretation of our results would be consistent with the results reported by Hon et al. (2007), who have shown that accounting for industry-specific effects is important for understanding the changes in the international comovement of stock returns that took place around March 2000 and for understanding the relevance of contagion effects. In order to illustrate our argument, it is worth noting that of the 340 firms that listed their stocks on the Neuer Markt between 1997 and 2000, 64% were operating in either the information and communications industry or in a related industry (Deutsche Börse (various issues)). Of the 160 firms that listed their stocks on the Nouveau Marché between 1996 and 2000, almost 70% were operating in this or a related industry (Bourse de Paris (various issues)). In contrast, firms listed on the AIM were less information-technology-oriented than those listed on the Neuer Markt and the Nouveau Marché. Of the 275 firms that listed their stocks on the AIM between 1998 and 2000, only 22% were operating in the IT industry or in a related industry (London Stock Exchange (various issues)). Thus, investors' beliefs about the return prospects of the IT industry might have been one reason why we found cross-Atlantic return linkages between the NASDAQ and the Neuer Markt and the Nouveau Marché, but not between the NASDAQ and the AIM in the pre-crash subsample. The difference between European stock markets for high-technology firms with regard to cross-Atlantic return comovement in the post-crash subsample might reflect structural differences between Germany, France, and the United Kingdom. According to a recent study by Beck and Levine (2002), the United Kingdom resembles the United States insofar as it is a leading market-based economy with high stock market capitalization. In contrast, France and Germany are both bank-based economies with comparatively low stock market capitalization. Thus, structural differences may have been one determinant of the strength of cross-Atlantic comovements in stock returns in stock markets for high-technology firms.