Modern Economy, 2011, 2, 569-574 doi:10.4236/me.2011.24063 Published Online September 2011 (http://www.SciRP.org/journal/me) Copyright © 2011 SciRes. ME Our Economy Christian Müller* Zurich University of Applied Sciences, School of Management and Law, Winterthur, Switzerland E-mail: much@zhaw.ch Received April 8, 2011; revised May 30, 2011; accepte d J une 12, 2011 Abstract I discuss the strengths and weaknesses of the current predominant approach to macroeconomic modelling of asset prices and suggest an alternative perspective. This alternative rests on the insight that the economy is the result of individual decisions. The industry standard has it, however, that individual action is ruled by objective, general laws instead. Changing the point of view allows to reconcile numerous puzzles and paves the way for a promising new research agenda. Keywords: Fundamental Uncertainty, Subjectivity, Finan cial Crisis 1. Introduction Many years ago the ancient Greek coined the term “eco- nomics” which defines the principle research agenda until today. Today we are still concerned with decision making for enhancing society’s welfare. However, two thousand years cannot pass leaving the world unchanged. While in ancient Greece decisions allocated the scarce resources of the oikos (the household) comprising the master, his family, slaves, and the land, today’s nomos (custom) of efficient action concerns thousands of households, countries and the world as a whole. The basic objective of economic analysis has never- theless remained largely unchanged. Economists are still looking for laws that can guide our behaviour to the bet- ter of the society. The most important difference—in my view—between then and now certainly is the degree of complexity of today's issues. In a relatively little world with well-defined roles for individuals, a judgment of the effects of one's action in the future appears reasonably reliable. We can thus imagine that the optimal decision can be made considering all relevant combinations of action and reaction. Interestingly, economists still apply the same strategy for advancing theories when describing human behaviour, or prescribing the best possible choice. The umbrella term for this procedure is usually called rational choice. Individuals are assumed to decide on the basis of rational expectations abou t the future state of the economy. In its probably most restrictive version th is approach posits the existence of the so-called homo oeconomicus. This hy- pothetical agent is, among other things, completely self- ish, profit oriented and processes an unlimited amount of information in no time. Not surprisingly, this methodo- logical approximation of real humans becomes more and more outdated as more and more evidence is gathered which proves the limits o f humans beyon d doub t. A main driver of this more recent development certainly are al- ternative analyses of decision making pioneered by re- searchers like [1,2], for example. Recognising the limits of the rationality concep t many authors have begun to consider plausible deviations from the rationality paradigm including (rational) learning, incomplete information and sentiments (see e.g. [3-5]) or ambiguity [6,7] to name but a few. Generally speaking, results derived by neighbouring sciences such as psy- chology which explain cognitive processes have entered economics and have helped to better understand individ- ual choice under more realistic assumptions [8,9]. Despite the significant augmentation of our under- standing of efficient decision making, numerous prob- lems remain. Many of them are known as puzzles be- cause the theoretical predictions are not matched by ac- cording observations. The puzzles I am referring to in particular are those in which the outcome of individual decisions seem not to be in line with considerations re- lated to the aggregate of information available. Popular examples are all sorts of exchange rate puzzles and asset price puzzles such as price bubbles. Typically, these puzzles feed on the seemingly mismatch of theoretically plausible prices and their actual counterparts. In the fol- lowing I will relate these puzzles to the predominant *I do thank participants of the 2009 Berlin--Copenhagen conference for many helpful comments. All mistakes are mine.
C. MÜLLER 570 concept of rationality. I then put the problem in the con- text of the traditional economists’ research program and finally, I will suggest a new research agenda. 2. The Dark Side of Rationality Recent research into the limits of rational behaviour can be compared to investigations of the moon by means of powerful telescopes. Prior to the invention of optical assistance the moon appeared more or less plain and bright whenever the sky was clear. The same was true for the concept of the rational agent. Ever since, however, we do know there are craters, mountains, and a whole lot of different structures on the moon’s surface, pretty much as there are serious scratches on the surface of the homo oeconomicus. After a while astronomers also learned that there was a dark side of the moon too, they would never be able to observe while looking from the earth. It was not until the first lunar explorers which left our planet that we finally got to kn ow the more complete story of the earth’s satellite. In my opinion, economists still await a similar en- deavour. What remains unknown to us is the “true” story behind economic agent’s decisions. This story is untold because we can only advance theories aiming at explain- ing behaviour but we can never know whether or not these theories reflect reality. Of course, there is empirical research which compares observations to what theories imply, but as mentioned earlier, key aspects of the economy such as exchange rate determination and asset pricing consistently defy satisfactory modelling giving rise to so-called puzzles. Why, then, can we compare these puzzles to lunar exploration? We can do so, be- cause, speaking allegorically, the main appro ach to solve those puzzles rests with the development of ever better telescopes. What we would really need, however, is a glance at the dark side of the moon. 2.1. Better Telescopes, or To understand this claim one may dissect the currently dominant empirical approach into two major parts. The first part comprises the collection of observations and their comparison to what has been expected on theoreti- cal grounds. This part alone provides enough issues for discussion to fill volume after volume of high ranking journals. Th e major source for the stream of publications has already been identified by [10] as follows. The re- searcher develops a model that uses (individual) expecta- tions about the future states on the economy. The corre- sponding empirical test can be carried out only conditional on the behav- ioural model . This means that conclusions concern- ing the expectations process will not be invariant to the choice of the underlying behavioural model [10, p.22]. Moreover, as almost always the choice of empirical data (definition, level of aggregation, transformations) itself is subject to discussion, researchers can regularly produce more insight based on variations of the model or the choice of the data. The choice of econometric tech- niques also nourish the publication stream. Directing a telescope towards the moon and taking notes is an as comparably transparent and competitive process of knowledge generation as this first part of the current ap- proach in economics. 2.2. Lunochod? The second constituent element of the current approach attracts far less attention, however. This element is the (tacit) assumption of the existence of an objective sto- chastic probability distribution of the future states of the economy. Notice, as theory defines a theoretic standard against which actual data is to be compared to, we must assume that such a means of comparison exists, and maybe more important, this standard must be independ- ent of the agents who are supposed to act on it. More precisely, objectivity is obtained by either imagining a representative agent or finite numbered groups of het- erogeneous agents. As the number of agents increases the total outcome tends towards some objective optimal de- cision which cannot be influenced by an individual. The formal condition is called ergodicity with respect to the number of individuals. For example, to render the fol- lowing equation meaningful from the point of view of an applied economist, its error term, [epsilon], has to follo w some ergodic stochastic process: =.yfx Here, y represents the observation, () the func- tional form and x all conditioning information that helps explaining y. Loosely speaking, the assumption of ergodicity with respect to the number of agents of any unexplained por- tion of y, that is , can be regarded the dark side of the moon. We would still not know the full story of our earth’s companion had we not escaped the gravity of convenience: the familiarity with the traditional telescope equipment. Likewise, unless we carefully scrutinise the implications of ergodicity, or rather non-ergodicity we are unlikely to fully appreciate human behaviour from an economist’s poi nt of vi ew. 3. Off Remote Control Let us reflect for a moment on the meaning of a repre- sentative agent. Any representative agent would find out Copyright © 2011 SciRes. ME
C. MÜLLER571 that y must equal x up to some agent specific mar- gin. Considering the repr esentative agent provides a han- dle for coping with the individual specific effect by as- suming that all individual effects follow some statistical law. In other words, there is a given probability that the individual specific margin does not exceed some upper and lower bound. Taking this assumption literally im- plies, however, that the subject is somehow ruled by a law that dominates his or her own will. Consequently, we cannot talk about subjects any more, the representative agent turns into an object. Considering heterogeneous agents instead does not change the principle as long as the degree of heterogeneity is finite. The representative agent approach yields a bizarre re- sult when applied to asset markets. Before turning to this issue let me first remind the reader that concepts like bubbles, and exuberance, and so on and so forth all tac- itly assume the existence of a correct, or true, or rational, or fundamental value of some financial asset. Unless we know this true price we cannot, h owever, attach the label “bubble” to prices exceeding this rational price, for ex- ample. According to the standard approach, any trader would agree on the price y. However, this price is deter- mined completely independent of the trader’s opinion. Therefore, when striking a deal this very trader is ulti- mately supposed to work on a kind of remote control. In my view, this is a totally misl eading, ev en b izarre picture of what is really going on. 4. The New Null Let me therefore suggest an alternative. This alternative starts with the simple observation that prices are set by humans. These subjects act on the basis of certain ex- ogenous conditions and their own will. Prices are thus the result of subjective judgements but not the outcome of some independent, objective process. Secondly, for a price to be quoted at least two subjects must interact. In 1836 David Ricardo already argued that a deal will only be beneficial when the parties involved are different, not identical. The same is true for asset prices. There is scope for a deal if the traders differ in their judgment about the perspectives of the future asset price. If we let the number of market participants increase, we should therefore expect that the degree of disagreement in- creases but not decreases like the representative agent approach has it. This increase in the degree of disagreement leads di- rectly to a new null hypothesis under which we may have a second look on existing empirical findings. If it was true that the number of agents in a market matters fo r the price process we should observe that the variance of the observed price increases the more agents are active. By contrast, the representative agent approach would sug- gest that the variance of the average price decreases. 5. Reconciliation If there ever was a chance to measure directly the rela- tionship between the number of agents and the variance of the asset price, for example, we would easily be able to reject or accept this hypothesis. However, there are considerable hassles to overcome because it is very dif- ficult to control for the number of people interacting. In my opinion, experiments are the potentially most pow- erful tools in that respect. Therefore, I scanned the exist- ing literature in order to look if a suitable experiment has ever been conducted. Unfortunately, I did not find any. What I found instead are numerous examples where it has been demonstrated how irrational price setting can become. So-called irrational behaviour can now be considered a stylised fact in artificial asset markets (see inter alia [1, 11]). It has also be demonstrated ([12-14]), however, that experienced traders can push the market price towards its fundamental value and hence eradicate irrational prices. Notably, all these experiments use a design where an (implicit) objective price process is induced. For exam- ple, the traded asset may yield a return with a given probability each period. Therefore, irrationality in such a situation might be used as an argument against the new null. I prefer a different interpretation, however. The participants in these experiment behave exactly as they would have done in the real world: they trade as if there was no objective price process. By contrast, expert trad- ers are able to discover the induced pricing rule and hence tend to behave rationally. Therefore, these ex- periments do not lend support to the standard approach. The decisive question is how do experts trade in the ab- sence of an objective price process? In sum, standard experiments, that is those in the vain of [1], use objective price processes whose very existence is hence not test- able. Because suitable experimental evidence is not (yet) available one may wonder if there are other bits and pieces of evidence for or against the new null hypothesis. In the following, I will focus on support while leaving the search for contradictions to future research. As long as direct tests of the relation between number of agents and variance have not been conducted we have to resort to various kinds of approximations. There are two kinds of approximation which I consider helpful. These are in the case of asset prices volume traded and size (length) of the order book. For example, [15] point out that the forward market is far less liquid than the spot market for foreign exchange to the effect that price volatility on the former is much smaller than on the latter market. There- fore, any regression of changes in spot rates on changes Copyright © 2011 SciRes. ME
C. MÜLLER 572 in forward rates (or forward-spot differences) can yield any result and is ultimately meaningless. Similarly, [16] report per-minute-data of the deutschmark—USD market where volume and volatility are clearly positively asso- ciated. The use of order book data has been popularised by [17]. These authors show that the regression fit of ex- change rate models increases dramatically when order book information is included. At the same time signifi- cance of “fundamentals” decreases considerably, or dis- appears completely. The regression fits the data better because the variance of the price is better captured. Us- ing the size of the orde r book as an approximation to the number of traders active allows us to understand this effect in terms of the subjective pricing process. In a further analysis, I have also run several regres- sions of share price volatility on the number of ticks per ten-minute-time interval as an approximation to the number of agents [18]. Again, the result is a clear posi- tive link between these two variables. Finally, I would like to hint to the notorious intra-day seasonality of financial market data. It is a well-estab- lished fact [19,20] th at share price volatility drops arou nd noon. This effect has been labelled lunch break puzzle. Under the new null hypothesis this puzzle disappears, however. The solution is straightforward: when traders have lunch they are not active on the market any more. We should therefore expect volatility to drop. It might be worth emphasising that the lunch break puzzle is so use- ful in support of the new null hypothesis because en- dogeneity bias is out of the question. Lunch is an exoge- nous event and any feedback from volatility to the oc- currence of lunch time can be safely excluded. 6. The Unexpectable In the light of the tentative evid ence let us operate under the assumption that the new null hypothesis holds and have a look at its implications. One implication affects the use of the term rationality of agents. If certain events are generic in the sense that subjective judgements de- termine their outcomes such as the pricing of assets, we would rationally conclude that an objective solution for determining these outcomes does not exist. Hence look- ing (only) in the direction of rational, representative agent models for predicting those outcomes becomes irrational itself. Secondly, because the outcome of human interaction on markets is ultimately subjective and not objective, objective probability distribution functions are of very limited use in general. Lack of any such probability dis- tribution implies that there are things which are not only very difficult to expect but which are even unexpectable. Is this conclusion worrisome? I don’t think so. First of all, life is life-threatening anyway. In othe r words, the future is open and no-one really knows, what it will bring. De- spite this fact the human race has been able to survive some one or two billions of years. Most of this time hu- mans were happy without the concept of rational deci- sion making in the modern economists' sense. Therefore, humans must have developed some tools for coping with the unexpectable which hence still await their discovery by economists. 7. Our Economy Economic agents somehow have to cope with the unex- pectable. These unexpectable events are in turn the re sult of the very subjects’ judgements and actions. In one word, the economy is shaped by ourselves and we do create the reality we live in ourselves. After all, it is our economy and nothing else. This statement is, of course in stark contrast to the many attempts of modelling ex- change rates, share prices, and the whole economy as chains of ev ents following objective prob ability distribu- tions. Provided the existence of the unexpectable we might wonder what the implication for economic analy- ses may be. In the following, I will raise some aspects I consider worth a more detailed investigation. First of all, I do not think that under the new null hy- pothesis representative agent models, or their ambigu- ity-augmented versions are totally useless or wrong al- together. The only adjustment that I deem necessary is the way we look at the respective findings and at what else we might be able to find. Let us reconsider the idea of the future being open. When things are fundamentally unexpectable while our day-to-day decisions are still based on guesses of what tomorrow will bring, we real- ize that individuals must use some mechanisms to either formulate those guesses or to find some other way to cope with the same problem. The principal agent ap- proach can thus be regarded as one single option out of a whole arsenal of weapons which arm us for coping with the unforeseeable. In this particular case, we try to ra- tionalise our actions and decisions to the greatest possi- ble extent on the basis of statistical analyses of past events. Once we take comfort in accepting the traditional ra- tional expectation—representative agent approach as one out of many possibilities for understanding economic decisions, the next obvious question is: what are the other tools? In my opinion, the search for these alterna- tives is the true challenge of future economic research. Luckily, economics has already become a very well di- versified science. Therefore, many mechanisms which help us making efficient use of resources when events are unexpectable have probably already been investi- gated. Those mechanisms could therefore simply been Copyright © 2011 SciRes. ME
C. MÜLLER573 reconsidered as part of the larger arsenal. There is no reason any more, however, to give priority to one par- ticular line of argument like rational expectations of a representative agent. In order to illustrate the last point consider the popular competition between so-called fundamentalists and so- called chartists. The traditional rational expectation ap- proach would clearly favour the analysis of fundamentals for explaining stock prices. If we take into account, however, that chartist create as much reality as do fun- damentalists, there is no reason to consider the analysis of fundamentals a priori more reasonable than the con- clusions drawn by chartists. Instead, we should search for arguments as to why either method is more suitable to handle the unexpectable; the answer to which is yet to be found. To provide another example, the thought that agents create the reality they have to deal with themselves brings back a number of issues economists have long regarded more or less settled. Consider for instance [21]’s case for flexible exchange rates. Six out of seven reasons Friedman gives for the determination of ex- change rates can directly be used as arguments in favour of flexible foreign exchange prices. The reasons are al- ways that exchange rates adjust to external, macroeco- nomic imbalances and vice versa which re-establish macroeconomic equilibrium. The assumption of such a bi-directional feedback mechanism is at the heart of many exchange rate puzzles, however. Under the new null hypothesis, a straightforward feedback from macro- economic conditions to exchange rates does no longer exist. Therefore, any short-cut to favouring flexible rates becomes doubtful. Again, there is no a priori reason ei- ther to jump to the conclusion that fixed rates are better. But we certainly have to consider the whole issue again under the new null hypothesis. 8. Points of Departure Returning to the subjectivity notion one might remember that psychologists have long noticed that decision mak- ing is a complex process which does not only involve those areas of the brain that are responsible for calcula- tion and thorough reasoning. In fact, humans can lose their ability to make a decision completely once the af- fective part of the brain is seriously damaged. It might very well be that the evolution has reserved a decisive role for emotions exactly because they equip us with the ability to cope with unexpectable events. Further relevant, seemingly irrational influence on de- cision making can be attributed to a tendency of neglect- ing information which runs against one’s initial convic- tions (see Section 5, third paragraph), using irrelevant “anchor” information, being more considerate when in a sad mood, being overwhelmed by too much information, and many more. Obviously, once we confront these be- havioural pattern with well-defined problems of stochas- tic optimisation we tend to find them utterly ridiculous. What we disregard in those comparisons, however, is the simple fact that well-defined problems rarely exist in real life. Therefore, economists should look at seemingly irrational patterns in the ligh t of un expectable events. Coming finally back to the initial example of an an- cient Greek household, one possible answer to the unex- pectable might have been given by societies which as- sign very well-defined roles to certain members of the society such as women, men, children, craftsmen, priests, aristocrats, and so on. Assigning these roles limits the possible outcomes of human interaction and hence re- duces the occurrence of unexpectable events. Therefore, such a strategy might benefit the society by making un- wanted unexpected events impossible. At the same time desirable unexpectable events such as economic devel- opment are also restricted. Hence, a classical trade-off results. If the outcome of a strategic choice is fundamen- tally uncertain, or unexpectable, non-classical tools for analysing this choice must be developed and applied. 9. Summary and Conclusions The outcome of human decisions and actions is as di- verse as humans are different from one another. There- fore, the outcomes are inherently subjective and system- atically defy mo delling by means of o bjective, stochastic processes. Observable human behaviour might hence be optimal in the sense that it is efficient given the occur- rence of unexpectable events while appearing at the same time “irrational” in laboratory settings. Traditional economic modelling can be regarded an approximation to the actual optimisation behaviour under the restriction that events are following some objective rules. However, to fully appreciate economic decision making we have to scrutinise individuals’ behaviour given that they also have to cope with the unexpectable. The results of these investigation will potentially yield important implications for policy making and theoretical research alike. 10. References [1] V. L. Smith, G. L. Suchanek and A. W. Williams, “Bub- bles, Crashes, and Endogenous Expectations in Experi- mental Spot Asset Markets,” Econometrica, Vol. 56, No. 5, September 1988, pp. 1119-1151. [2] D. Kahneman and A. Tversky, “Prospect Theory: An Analysis of Decision under Risk,” Econometrica, Vol. 47, No. 2, 1979, pp. 263-291. [3] P. De Grauwe and P. R. Kaltwasser, “Modeling Opti- Copyright © 2011 SciRes. ME
C. MÜLLER Copyright © 2011 SciRes. ME 574 mism and Pessimism in the Foreign Exchange Market,” CESifo Working Paper Series 1962, CESifo GmbH, April 2007. [4] P. Bacchetta and E. van Wincoop, “Rational Inattention: Solution to the Forward Discount Puzzle,” Research Pa- per 156, International Center for Financial Asset and En- gineering, September 2005. [5] C. Sims, “Rational Inattention: A Research Agenda,” Discussion Paper, Series 1: Economic Studies 34, Deutsche Bundesbank, 2005. [6] J. W. Milnor, “Games against Nature,” In: C. H. Coombs, R. L. Davis and Robert McDowell Thrall, Eds., Decision Processes, Wiley, New York, 1954, pp. 49-60. [7] E. Hanany and P. Klibanoff, “Updating Ambiguity Averse Preferences,” The B.E. Journal of Theoretical Economics, Vol. 9, No. 1, 2009, p. 37. [8] J. Conlisk, “Why Bounded Rationality?” Journal of Eco- nomic Literature, Vol. 34, No. 2, June 1996, pp. 669-700. [9] J. Tirole, “Rational Irrationality: Some Economics of Self-Management,” European Economic Review, Vol. 46, No. 4-5, May 2002, pp. 633-655. [10] H. M. Pesaran, “The Limits to Rational Expectations,” Basil Blackwell, Oxford, 1987. [11] M. Cipriani and A. Guarino, “Herd Behavior in a Labo- ratory Financial Market,” American Economic Review, Vol. 95, No. 5, December 2005, pp. 1427-1443. doi:10.1257/000282805775014443 [12] M. Dufwenberg, T. Lindqvist and E. Moore, “Bubbles and Experience: An Experiment,” American Economic Review, Vol. 95, No. 5, December 2005, pp. 1731-1737. doi:10.1257/000282805775014362 [13] M. Drehmann, J. Oechsler and A. Roider, “Herding and Contrarian Behavior in Financial Markets: An Internet Experiment,” American Economic Review, Vol. 95, No. 5, December 2005, pp. 1403-1426. doi:10.1257/000282805775014317 [14] R. N. Hussam, D. Porter and V. L. Smith, “Thar She Blows: Can Bubbles Be Rekindled with Experienced Subjects?” American Economic Review, Vol. 98, No. 3, June 2008, pp. 924-937. doi:10.1257/aer.98.3.924 [15] P. Wang and T. Jones, “The Impossibility of Meaningful Efficient Market Parameters in Testing for the Spot- Foreward Relationship in Foreign Exchange Markets,” Economics Letters, Vol. 81, 2003, pp. 81-87. doi:10.1016/S0165-1765(03)00148-4 [16] J. A. Carlson and M. Lo, “One Minute in the Life of the DM/US$: Public News in an Electronic Market,” Journal of International Money and Finance, Vol. 25, 2006, pp. 109-1102.doi:10.1016/j.jimonfin.2006.08.005 [17] M. D. D. Evans and R. K. Lyons, “Order Flow and Ex- change Rate Dynamics,” Journal of Political Economy, Vol. 110, No. 1, February 2002, pp. 170-180. doi:10.1086/324391 [18] C. Mueller-Kademann, “Puzzle Solver,” MPRA Paper 19852, University Library of Munich, Germany, October 2009. [19] T. Ito, R. K. Lyons and M. T. Melvin, “Is There Private Information in the FX Market? The Tokyo Experiment,” Journal of Finance, Vol. 53, No. 3, 1998, pp. 1111-1130. [20] P. Hartmann, M. Manna and A. Manzanares, “The Mi- crostructure of the Euro Money Market,” Journal of In- ternational Money and Finance, Vol. 20, No. 6, Novem- ber 2001, pp. 895-948. doi:10.1016/S0261-5606(01)00029-8 [21] M. Friedman, “The Case of Flexible Exchange Rated,” In: M. Friedman, Ed., Essays in Positive Economics, Univer- sity of Chicago Press, Chicago, 1953, pp. 157-203.
|