کارایی بازار و آموزش در بازار سهام مصنوعی: چشم انداز از اقتصاد نئو اتریش
|کد مقاله||سال انتشار||تعداد صفحات مقاله انگلیسی||ترجمه فارسی|
|13344||2010||21 صفحه PDF||سفارش دهید|
نسخه انگلیسی مقاله همین الان قابل دانلود است.
هزینه ترجمه مقاله بر اساس تعداد کلمات مقاله انگلیسی محاسبه می شود.
این مقاله تقریباً شامل 13689 کلمه می باشد.
هزینه ترجمه مقاله توسط مترجمان با تجربه، طبق جدول زیر محاسبه می شود:
- تولید محتوا با مقالات ISI برای سایت یا وبلاگ شما
- تولید محتوا با مقالات ISI برای کتاب شما
- تولید محتوا با مقالات ISI برای نشریه یا رسانه شما
پیشنهاد می کنیم کیفیت محتوای سایت خود را با استفاده از منابع علمی، افزایش دهید.
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Empirical Finance, Volume 17, Issue 4, September 2010, Pages 668–688
An agent-based artificial financial market (AFM) is used to study market efficiency and learning in the context of the Neo-Austrian economic paradigm. Efficiency is defined in terms of the “excess” profits associated with different trading strategies, where excess is defined relative to a dynamic buy and hold benchmark in order to make a clean separation between trading gains and market gains. We define an Inefficiency matrix that takes into account the difference in excess profits of one trading strategy versus another (signal) relative to the standard error of those profits (noise) and use this statistical measure to gauge the degree of market efficiency. A one-parameter family of trading strategies is considered, the value of the parameter measuring the relative informational advantage of one strategy versus another. Efficiency is then investigated in terms of the composition of the market defined in terms of the relative proportions of traders using a particular strategy and the parameter values associated with the strategies. We show that markets are more efficient when informational advantages are small (small signal) and when there are many coexisting signals. Learning is introduced by considering “copycat” traders that learn the relative values of the different strategies in the market and copy the most successful one. We show how such learning leads to a more informationally efficient market but can also lead to a less efficient market as measured in terms of excess profits. It is also shown how the presence of exogeneous information shocks that change trader expectations increases efficiency and complicates the inference problem of copycats.
In recent years it has become ever more popular to consider financial markets from other than a neoclassical rational expectations point of view. The latter considers financial markets to be in continuous equilibrium with informationally efficient prices. Empiricists have questioned the validity of this model, pointing to evidence of inefficiencies. Alternative views have been presented to better match the empirical evidence. One with a distinguished history, that will be the focus of this paper, is the Neo-Austrian theory of financial markets. Based on a recent rereading of the ideas of Friedrich Hayek and the Neo-Austrian theory of market processes (see, e.g., Hayek, 1937, Hayek, 1945, Hayek, 1948, Hayek, 1978, Littlechild, 1982, Rizzo, 1990, Kirzner, 1992, Kirzner, 1997 and Benink and Bossaerts, 2001), Benink and Bossaerts (2001) presented the first formal application of Neo-Austrian theory to financial markets. In the Neo-Austrian interpretation financial markets are continuously evolving from one inefficiency to another, never attaining the perfect, efficient equilibrium, yet strongly attracted towards it. Creative investors track and exploit profit opportunities generated by continuous shocks in a never-ending cycle. The result would be a stable process with pronounced regularities. According to Neo-Austrian theory, a competitive market provides a systematic set of forces, put in motion by entrepreneurial alertness (i.e. eagerness to make money), which tend to reduce the extent of ignorance among market participants. The resulting knowledge is not perfect; neither is ignorance necessarily invincible. Equilibrium–read: market efficiency–is never attained, yet the market does exhibit powerful tendencies towards it. The fact that equilibrium is never attained is attributed to an erratically changing world where traders realize that their knowledge is imperfect. At the same time, the changes are never so extreme as to frustrate the emergence of powerful and pervasive economic regularities. Although traders can exhibit fully rational behaviour, in the sense that they try to optimize their financial position and wealth, the market process is not generating a rational expectations equilibrium (REE) and informationally efficient prices. Rational behaviour does not necessarily imply rational expectations.1 Imperfect knowledge is a key characteristic of Neo-Austrian thinking. According to Hayek, the problem of economic choice and ultimately the analysis of economic behaviour in neoclassical theory is oversimplified, because it has been reduced to optimal behaviour under constraints that agents are supposed to be very familiar with. These constraints concern: (1) preferences, (2) production and market technology, and (3) resources. In contrast, the Neo-Austrian view stresses that fundamental uncertainty and ignorance exist regarding these constraints. This uncertainty and ignorance is claimed to lead to disequilibrium, and disequilibrium itself generates further uncertainty and ignorance regarding the constraints. Disequilibrium thereby becomes self-enforcing and permanent. However, alert participants in the market process, whom the Neo-Austrians define as entrepreneurs, try to get a–necessarily incomplete–picture of the nature of the disequilibrium in the marketplace, because disequilibrium generates profit opportunities. The actions of these entrepreneurs produce the very signals that are needed to reduce disequilibrium. What renders the market process a systematic process of coordination is the circumstance that each gap in market coordination expresses itself as a pure profit opportunity. The profit-grasping actions of successful entrepreneurs dispel the ignorance which was responsible for the profit opportunities, and thus generate a tendency towards coordination among market decisions. However, due to continuous change in the constraints, equilibrium is never reached. In their paper Benink and Bossaerts (2001) construct an example of an economy with a continuously inefficient financial market. They adjust the memory of investors' trading rules in order to generate a market process that can be characterized as stable, cycling from one inefficiency to another. Despite the stability (stationarity), rational, risk-averse investors are unable to exploit all inefficiencies because they cannot make reliable inferences about them. This would be the case, for instance, if the memory of the return process is sufficiently long for statistics not to display their usual distributional properties needed to construct confidence intervals. Based on an analysis of average price changes, an investor will with high likelihood reject efficiency, yet the sign of the average is unreliable in predicting the sign in independent future replication. As a consequence, classical statistics cannot reliably assess the inefficiencies. As a follow-up to Benink and Bossaerts, this paper places more emphasis on, and studies in detail, the learning processes and dynamics of a Neo-Austrian inefficient financial market. As mentioned above, the neoclassical rational expectations point of view considers financial markets to be in continuous equilibrium with informationally efficient prices. Pesaran (1989) notes that the idea of a REE involves much more than the familiar concept of the equilibrium of demand and supply. A REE can be characterized by three main features: (1) all markets clear at equilibrium prices, (2) every agent knows the relationship between equilibrium prices and private information of all other agents, and (3) the information contained in equilibrium prices is fully exploited by all agents in making inferences about the private information of others. Thus, in a REE prices perform a dual role — apart from clearing the markets they also reveal to every agent the private information of all the other agents. In effect, the concept of the REE requires that everybody knows (in a probabilistic sense) everything about the way the market economy functions. But as Hayek (1937) puts it: “The statement that, if people know everything, they are in equilibrium is true simply because that is how we define equilibrium. The assumption of a perfect market in that sense is just another way of saying that equilibrium exists, but does not get us any nearer an explanation of when and how such a state will come about. It is clear that if we want to make the assertion that under certain conditions people will approach that state we must explain by what process they will acquire the necessary knowledge”. The preceding implies that, for the REE to have any operational meaning, it is necessary that the processes by means of which people learn from experience and acquire the common knowledge necessary for the achievement of the REE, are specified fully and explicitly.2 In this paper we use an agent-based artificial financial market (AFM) to generate simulations of inefficiencies and learning and investigate to what extent a Neo-Austrian interpretation of the resulting market dynamics is the most natural.3 Agent-based models are intermediate between empirical and analytic studies; the former offering grave problems in terms of inference, while the latter, perforce, come armed with a large number of model assumptions. Moreover, the complexity of the AFM can be tuned, so as to offer a more transparent model versus a more realistic one where it is difficult to understand the relations between inputs and outputs when market parameters are changed. The AFM we deal with is deliberately kept simple as our primary concern is to be able to intuitively understand how the behaviour of the AFM changes while changing parameters of the AFM (such as the trading strategies).4 The most well known AFM is the Santa Fe model (Palmer et al., 1994, Arthur et al., 1997, LeBaron, 1999, LeBaron, 2000 and LeBaron, 2001). In this paper we use an alternative model–the so-called Neural Networks Chaos and Prediction Model (NNCP) (Gordillo and Stephens, 2001a and Gordillo and Stephens, 2001b)–whose design was motivated by the desire to study relatively neglected elements, such as the effect of organizational structure on market dynamics and the role of market makers and information, all of which are important in the formation of market microstructure (see, for example O'Hara (1997)). Although capable of simulating more “realistic” dynamical scenarios, in this paper we use the NNCP in the context of a more transparent model, in which traders are associated with trading strategies chosen from a single one-parameter family, the parameter representing a trading bias linked to the informational advantage of the trader, zero bias representing noise traders. The resulting AFM, presented in Section 2, can effectively be parameterized by three principle degrees of freedom: (1) the proportion of traders of a given type, (2) the number of different trader groups or strategies, and (3) the similarity between different trader groups — measured by distance in bias between two agents or groups. Learning is introduced in Section 2.1 via the notion of “copycat” agents that observe the market, infer what is the most successful strategy and then copy it. We use this AFM to investigate notions of efficiency and learning and examine to what extent the results are more naturally interpretable in a Neo-Austrian rather than a neo-classical framework. AFMs have been used, for example, by Chen and Yeh (2002), to consider efficiency as an emergent phenomenon. There however, efficiency was judged purely from the statistical properties of the returns series. However, as will be further discussed in Section 3, predictability of the time series is not necessarily inconsistent with market efficiency. We therefore consider efficiency from the empirical point of view of whether or not traders can make excess returns systematically, defining a notion of excess profit that distinguishes between market gains and trading gains. To further distinguish between intelligent trading and “luck” we consider, following Benink et al., 2004 and Stephens et al., 2007, relative excess returns, Iij, between trading strategies i and j, measured relative to the variance of these excess returns. In Section 3, we introduce an Inefficiency Matrix, with matrix elements Iij, which summarizes statistically the relevant relative inefficiencies in the market. With these tools in hand, in 1 and 2, we investigate both efficient and inefficient markets in the absence of learning, showing in particular, in Section 4.1, under what conditions a market may be inefficient, yet still be observed to be efficient. This possibility is due to the statistical inference problem that traders face in the light of noisy market data. In Section 4.3 we show, paradoxically, that learning can lead to a more inefficient market in terms of excess profits, even though informationally the market was more efficient; and then, in Section 4.4, we study how the arrival of new information affects inefficiency and learning. Finally, in Section 5, we discuss the results in the framework of the Neo-Austrian paradigm and draw some final conclusions.
نتیجه گیری انگلیسی
The main goal of this paper was to study Neo-Austrian ideas about market efficiency and learning, usually expressed in qualitative language, in the formal setting of an agent-based AFM market — the NNCP. Unlike a real market, the luxury of an AFM is that we can create an efficient or inefficient market and then vary the parameters of the market in order to study when, and under what conditions, it is possible to infer efficiency from observing the market. To avoid the joint-hypothesis problem we used a purely empirical quantitative measure of efficiency, defining an inefficient market in terms of whether or not there exist traders making systematic, excess profits. To distinguish between trading gains and market gains excess was defined relative to a dynamic buy and hold benchmark. As excess profit is a stochastic variable it is most naturally measured in units of the standard deviation of the excess profits. In this way one can not only distinguish between traders profiting from an active trading strategy as opposed to those profiting purely from market gains, but one can also distinguish between those traders who have a “lucky” trading strategy versus an “intelligent” one. We used the concept of an Inefficiency Matrix which summarizes the relative inefficiencies between the different trading strategies in the market. Using the Inefficiency Matrix as the principle measure, we performed a series of simulations in the context of a transparent model where the traders used trading strategies taken from a one-parameter family associated with a trading “bias”, zero bias corresponding to noise or liquidity traders. Learning was introduced using the concept of a copycat trader who observes the market, statistically estimates which is the most successful strategy and then copies it. It is important to emphasize that all these elements have been used in previous studies, in Gordillo and Stephens, 2001a and Gordillo and Stephens, 2001b, without any reference to the Neo-Austrian paradigm. In other words, our model was not designed with the Neo-Austrian viewpoint in mind. Rather, our motivation was to use a previously designed model to see to what extent its results were most naturally interpreted in the Neo-Austrian as opposed to the neoclassical framework. Two important observations gleaned from these simulations are: Firstly, that market inefficiency, as we have defined it, perforce requires the presence of heterogeneity, i.e. distinct trading strategies. Markets where all participants utilise the same trading strategy are incapable of exhibiting inefficiencies. Secondly, that efficient learning in the context of the copycat paradigm requires that the different trading strategies be readily distinguishable in terms of their relative performance, as, otherwise, agents can easily make mistakes in the learning process as to what is the optimal strategy to follow. Allied to this is the fact that the learning process itself can naturally lead to a greater convergence of trading strategies, as copycats switch to the more successful ones. This negative feedback process is, in fact, the market becoming more efficient. The results of this paper are in general consistent with the Neo-Austrian interpretation of markets as opposed to the neoclassical rational expectations point of view that considers financial markets to be in continuous equilibrium with informationally efficient prices. The results, in fact, lend substantial insight into and enrich many of the key elements of Neo-Austrian theory which, as mentioned, is a predominantly qualitative theory. For instance, we saw that the existence of alert entrepreneurs (copycats) could even lead to an increase in inefficiency rather than a decrease, depending on the proportion of informed to uninformed agents in the market. Interestingly, and contrary to other alternative theories explaining inefficient financial markets, in the Neo-Austrian view the failure of markets to reach the informationally efficient equilibrium ought not to be attributed to costs of any nature (adjustment costs, information costs, trading costs, etc). This description fits very well the results of the simulations of Section 4.4, where “creative investors” (informed traders and copycats) track and exploit profit opportunities generated by exogeneous information shocks. With their emphasis on imperfect knowledge the Neo-Austrians put themselves at the heart of the famous debate on risk and uncertainty (see, e.g., Knight (1921)). Neoclassical financial economists believe that uncertainty can be reduced to “objective” risk, depending on knowledge of the objective probability distribution implied by the true model of the economy that economic agents know or are capable of learning. However, Neo-Austrians tend to emphasize that economic agents have to cope with imperfect knowledge and fundamental uncertainty. Just as post-Keynesian economists, they claim that neoclassical theory fails to specify how agents will be able to overcome fully this uncertainty, i.e. that it can be reduced to the “objective” probability distribution implying rational expectations and efficient markets. Contrary to post-Keynesians, however, Neo-Austrians claim that alert market participants have powerful incentives to learn about the true nature of uncertainty and related disequilibrium, because disequilibrium generates profit opportunities. Thus, the market process is viewed as a stabilizing process with a powerful tendency towards equilibrium and efficiency. With their trust in the market process Neo-Austrian economists are intellectually close to their neoclassical colleagues, although they arrive at this result from a rather different perspective on the underlying economic reality. We believe that our results on learning and inference for copycat entrepreneurs perfectly illustrate this point of view. Due to their flexibility and adjustable complexity, we believe that AFMs are an ideal vehicle for addressing some of the deepest and most difficult questions about efficiency and rational expectations. Further, by using a purely empirical measure of inefficiency, as we have done here, complications due to the joint-hypothesis problem can be avoided. We believe that combining the two gives a powerful framework within which other fruitful studies may be carried out.