مدل سازی ارتباط های منطقه ای بازارهای مالی
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|14158||2014||14 صفحه PDF||سفارش دهید||7644 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Economic Behavior & Organization, Volume 99, March 2014, Pages 18–31
With the development of globalization and regional economic integration, regional markets linked with a common currency emerge, in which investors from domestic market are allowed to trade in foreign markets. Empirical studies have evidenced extensively the existence of co-movement of asset prices or cross-correlation in market returns among these markets, especially in global event. However, there is no theoretical model in literature that can provide economically plausible justifications for these stylized facts. This research intends to fill up such a gap with a simplest possible nonlinear dynamic model. Based on the classical market-maker framework of Day and Huang (1990), a two-market HAM model is developed, which does not only prove in theory the existence of price co-movement but also replicate in simulation this typical characteristic, along with other well known stylized facts characterizing individual financial market. Moreover, theoretical analysis suggests meaningful implications for market opening policy. In particular, in terms of financial stability, a relatively small market may not benefit from market linkage and market opening is essentially a double-edged sword.
With the development of regional market integration, linkage among markets become stronger and common currency circulating within the region emerges. A typical example is Euro, which serves as the transaction currency for all financial markets within the euro-zone. Examples of regional asset markets include but are not limited to the Shanghai and Shenzhen stock exchange markets in China, and NASDAQ and New York stock exchanges in United States. The common transaction currency eliminates the need of currencies exchange so as to remove the impacts from the foreign exchange market. Along with financial market integration comes markets co-movement or cross-correlation, as it has been widely reported in empirical literature. Kenett et al. (2012b) find that developed Western markets are highly correlated. Egert and Kocenda (2011) find strong correlation among returns of Germany, France and UK, even up to 0.9. In addition, strong co-movement was observed when financial crisis spread to many markets in the past decades. Preis et al. (2012) show that average correlation among DJIA members increases with market stress. Kenett et al. (2012a) report that interdependencies between markets increase at times of global events. Regrettably, there is no theoretical model in literature that can simulate the prices with these stylized facts. Heterogeneous agents models (HAM) have so far managed to calibrate successfully some of the financial market stylized facts related to individual financial market, among which are large trading volume, cluster volatility, returns distribution with fat tails, and unpredictable asset returns with almost no autocorrelation. For reference, we cite Huang and Day, 1993, Lux, 1995, Lux, 1998, Lux and Marchesi, 2000 and Brock and Lebaron, 1996 and Farmer and Joshi (2002). There is a call to build a multi-market model with capability of not only replicating the stylized facts of prices co-movement, but also offering economically plausible explanations. Such model can enhance our understanding of the integrated financial system and further shed light on the study of the propagation mechanism of financial crisis. This research intends to fill up such a gap by building a simplest possible nonlinear dynamic HAM model. For this purpose, a market system composed of two markets linked by a common transaction currency is studied so that the investors are allowed to invest in both markets with no hassle of exchange rate. Such set up is shown to be able to replicate the typical characteristics for multi-markets system in additional to other well known stylized facts characterizing individual financial market. Day and Huang (1990) introduce a stylized market maker framework in which two agent types, chartist and fundamentalist, invest in an asset market and a market maker updates price in each period. The model is in discrete time and exhibits complicated, chaotic price fluctuations around a fundamental price with random switching between bear and bull market episodes. Instead of the market maker framework, Brock and Hommes (1998) apply the Walrasian equilibrium concept in heterogeneous agents model. In their model, micro-foundation is built on a fitness measure, which is determined by the past realized profit. Every period, agent composition is determined by fitness measures. Agent aims to maximize investment profit and decides her supply and demand according to the chosen strategy. Market clears at the end of each period. This model is capable of explaining some stylized financial behaviors such as irregular switching among phases of price movements. However, LeBaron (2006) argue that the market clearing Walrasian equilibrium in every period has limitations. One of the limitations is that it may not represent the continuous trading of financial market accurately. Nevertheless, the combination of market maker and micro-foundation based on fitness measure develops in later literature such as Westerhoff, 2004, He and Westerhoff, 2005, Westerhoff and Dieci, 2006 and He and Li, 2008 and Huang et al. (2010). Majority of the heterogeneous agents models focus on a single market or one risky asset with reference to one riskless asset. Recently, the idea of heterogeneous agents is extended to price dynamics of multi-asset within a market, or even to the interactional dynamics of multi-markets. For example, Bohm and Wenzelburger (2005) investigate the performance of efficient portfolios in a financial market in which heterogeneous investors including rational traders, noise traders, and chartists are active. Brock et al. (2009) introduce additional Arrow securities into the stylized evolutionary equilibrium model of Brock and Hommes (1998) and demonstrate that more hedging instruments may destabilize markets with heterogeneous agents and performance-based reinforcement learning. Westerhoff and Dieci (2006) develop a model in which chartists and fundamentalists invest in two speculative markets. The composition of investors varies according to profit fitness measurement. After stability conditions for the fundamental steady state are derived, the model generates complex price dynamics resembling to actual speculative prices. Dieci and Westerhoff (2010) build up a three-market model in which two stock markets are linked via foreign exchange market. The foreign exchange market is populated with chartists and fundamentalists while the two stock markets have only fundamentalists. It is concluded that upon market interactions, stock markets may be destabilized while the stabilizing effect on the foreign exchange market and the whole market system can be observed. This paper follows the framework of Day and Huang (1990) and Westerhoff and Dieci (2006). Two types of investors, chartists and fundamentalists, invest in two speculative markets with the same transaction currency. Each investor can invest in either market and choose a chartist or fundamentalist strategy in each market. The difference in this paper is that agents of the same type are inhomogeneous across markets. That is, chartists or fundamentalists from different markets have different demand strengths. In addition, factor of market size/population is included to investigate its role. Theoretical analysis and simulations show that a market that is more stable initially will stabilize the market system while it is subjected to destabilizing effect from the market system. This mutual effect also applies to a market that is more unstable initially. Interpreting from the population size of individual market, a market with a smaller population has lesser influence on the market linkage. This paper is structured as follows. For the purpose of comparison, we start with in Section 2 a hypothesized case in which two regional markets are isolated with each other in the sense that the investors are not allowed to invest in the foreign market. Section 3 then explores the case when these two isolated markets are linked by allowing the investors from each market to invest in both. Theoretical analysis is carried out so that meaningful policy implications can be drawn. Section 4 provides various numerical simulations and verifies its capability to generate price series matching typical stylized facts documented in the literature, especially the price co-movement and the cross-correlation in volatility. Section 5 concludes with the directions of future research. 2. Market isolation Following the market maker framework of Day and Huang (1990), we assume that a financial market is composed of three types of agents: chartists, fundamentalists and a market maker. Fundamentalists behave in a way that they sell over-priced asset and purchase under-priced one. In contrast, chartists simply assume the persistence of bullish and bearish market episodes in the short run. Following this expectation, they purchase the over-priced asset and sell the under-valued one. There exist two regional stock markets, denoted by A and B. We first assume that these two markets are isolated so that investors are allowed to invest in their domestic market only. The composition of chartists and fundamentalists among the investors depends on market circumstance. Fundamentalists play roles of correcting market price and their composition would become larger for a larger price deviation, as demonstrated by Hommes (2001). A larger asset price deviation triggers more agents to rely on fundamentalist strategy based on the micro-foundation of fitness measures.
نتیجه گیری انگلیسی
A two-market heterogeneous agents model is developed. Each market has a market maker and two types of agents: chartists and fundamentalists. Agents of the same type are inhomogeneous across markets. Market linkage is established by allowing investors to invest in each market. Aware of capital movement of the investors and common factors underlying the two markets, individual market maker updates price for her market by weighing excess demands of both markets to take account of the impact from the other market. Existence of price co-movement/cross-correlation between markets is proved. By establishing market linkage, individual market's intrinsic dynamic properties may be overwritten. A market that is more stable initially in isolation will exert stabilizing effect on the market system while it will be subjected to destabilizing effect from the resultant market system. In addition, a market with a larger population has larger influence over the resulted assets prices of the market system. This market linkage can provide policy implication for financial market opening. In a world consisting of a small market and a large market (or market agglomeration), if the small market is stable compared to the large market, market opening of the small market will cause the small market to destabilize. Small market will benefit from market opening only if it is unstable originally compared to the large market. This example indicates that market opening is a double-edged sword. Decision of market opening should be based on the impact assessment on internal and external markets. Lastly, numerical simulations demonstrate the model's capability to generate some of the stylized facts of speculative financial market, especially the cross-correlation between markets. To our best knowledge, this is the first HAM model that is capable of generating the significant cross-correlation effect. This can be useful to study multi-market financial system as cross-correlation should become more and more evident given the current trend of financial integration. With market opening and financial market integration, investors enjoy lower transaction cost and more investment opportunities. However, we should be aware of the other side of the coin. If the process of financial market integration is inevitable, more and more markets tend to synchronize and market coupling and cross-correlation between markets should become larger, especially during global events such as financial crisis. In that extreme case, we will observe financial crisis contagion spreading from one market to another. Further studies of market cross-correlation are urged to have an insightful understanding of market interactions. We cite a latest study Huang and Chen (2013). They study market-system with more than two markets and demonstrate the formation of market clusters with same direction of asset price deviation. For financial crisis, the existing HAM literatures focus on a single market, e.g. Huang and Zheng (2012) and Huang et al. (2013). Our two-market model can be applied to study multi-market financial crisis and shed some light on multiple markets interactions. As cross-correlation is the key point of market interaction, another direction of future research can be behavioral explanation for market correlation from aspect of market data rather than statistical fitting, as exemplified by Feng et al. (2012).