Behavioral Economics: Reunifying Psychology and Economics, Apuntes de Fisioterapia

¡Descarga Behavioral Economics: Reunifying Psychology and Economics y más Apuntes en PDF de Fisioterapia solo en Docsity!

M PRA

Munich Personal RePEc Archive

Behavioral Economics

Nathan Berg

Online at http://mpra.ub.uni-muenchen.de/26587/

MPRA Paper No. 26587, posted 2. December 2010 09:59 UTC

Behavioral Economics

Nathan Berg [email protected] University of Texas-Dallas

Introduction: Definitions and Naming Problems in Behavioral Economics

Definition of Behavioral Economics Behavioral economics is the subfield of economics that borrows from psychology, empirically tests assumptions used elsewhere in economics, and provides theories that aim to be more realistic and closely tied to experimental and field data. In a frequently cited survey article, Rabin (1998) describes behavioral economics as ―psychology and economics,‖ which is a frequently used synonym for behavioral economics. Similarly, Camerer (1999) defines behavioral economics as a research program aimed at reunifying psychology and economics. Reunification is a relevant description because of the rather tumultuous relationship between psychology and economics in the arc of economic history. A number of preeminent founders of important schools of economic thought, including Adam Smith, wrote extensively on psychological dimensions of human experience and economic behavior, while later economists sometimes sought explicitly to exclude psychology from economic analysis. For example, Slutsky, whose famous equation is taught to nearly all upper-level microeconomics students, sought to erect a boundary excluding psychology from economics: ―[I]f we wish to place

observers to see more similarity than contrast in the behavioral and neoclassical approaches (Berg and Gigerenzer, forthcoming).

Bounded Rationality The term bounded rationality, coined by Nobel laureate, Herbert Simon, is strongly associated with behavioral economics, although there appears to be far less agreement on the term‘s meaning. The neoclassical model assumes that economic man, or homo economicus , is infinitely self-interested, infinitely capable of processing information and solving optimization problems, and infinitely self-disciplined or self-consistent when it comes to having the willpower to execute one‘s plans—whether those plans concern how much junk food to eat or how much to save for retirement. In contrast, much of behavioral economics focuses on limits, or bounds, on one or more of these three assumptions. Thus, bounded self-interest, bounded information processing capacity, and bounded willpower are three guiding themes in the behavioral economics literature. Bounded self-interest enjoys widespread interest in behavioral economics, which has proposed numerous models of so-called social preferences to address a number of observations from human experiments that appear to falsify the assumption that people maximize their own monetary payoffs. A decision maker with social preferences cares about the material or monetary payoffs of others as well as his or her own, although the manner in which concern for others‘ payoffs is expressed can take a variety of forms. For example, a person with social preferences might be happier when others are worse off, which is sometimes described as spite; happier when others are better off, which is sometimes described as altruism; prefer equal over unequal allocations of money, which is sometimes described as inequality aversion; prefer

allocations in which the sum of all people‘s payoffs is maximized, which is sometimes described as a preference for social welfare; prefer allocations in which the least well-off person has a larger payoff, which is sometimes described as a Rawlsian preference; or prefer allocations of resources in which his or her payoff is large relative to others, which is sometimes described as a competitive preference (Charness and Grosskopf, 2001). Common to all these variations and many other forms of social preferences is that people are not generally indifferent between two allocations of payoffs for all members in a group just because their own monetary payoff is the same. This violates the common neoclassical assumption that people are infinitely self- interested, because it would imply that people are indifferent so long as their own material payoffs are held constant. Bounded information-processing capacity is another active area within behavioral economics, which would have looked very much out of place in the mainstream economics literature only three decades ago. Topics in this area include limited memory, limited attention, limited number of degrees of perspective-taking in strategic interaction, limited perceptual capacity, distorted beliefs, and decision and inference processes that violate various tenets of logic and probability theory (see Camerer, 2003, for examples). Bounded willpower, often described as time-inconsistency or dynamic inconsistency, is another large and growing part of the behavioral economics research program. In the neoclassical optimization model, decision makers choose a sequence of actions through time by selecting the best feasible sequence, with virtually no mention of the costs associated with implementing that plan over the course of people‘s lives. If there is no new information, then the neoclassical inter-temporal choice problem is decided once and for all before the first action in the sequence is taken. Unlike the neoclassical model‘s assumption that acting on the optimal

Example of reference-point dependent utility functions Similarly to Simon, Rabin (1998) argues that theories and experimental results from psychology enrich mainstream economics. But Rabin‘s idea about how behavioral economists can bring more empirical realism into economics is much more narrowly circumscribed than Simon‘s, with Rabin essentially arguing that behavioral economics should proceed within the utility maximization model of neoclassical economics. Despite their occasional claims to radical or revolutionary methodological innovation, many behavioral economists side with neoclassical economists in viewing constrained optimization as a non-negotiable methodological tenet that defines and distinguishes economics from other disciplines. Rabin says that the new empirical content that behavioral economists bring to bear will help economics as a whole to more realistically describe people‘s utility functions. For example, as mentioned earlier in the discussion of the neoclassical model‘s assumption of unbounded self-interest, this assumption is often interpreted to mean that consumers care only about their own levels of consumption and that workers care only about their own income, irrespective of what others are consuming or earning. Behavioral economics models, in contrast, allow utility to depend on the difference between one‘s own level of consumption or income and a reference-point level. The reference-point level might reflect what one is accustomed to or reflect a social comparison made with respect to the average level within a social group. Thus, a worker with a behavioral reference-point-dependent utility function might prefer an annual salary of $90,000 at a company where the average worker earns $50, over a salary of $95,000 at a company where the average worker earns $200,000. In the

standard economic model, only the worker‘s own payoffs should determine the ranking of job opportunities, holding all else equal, and not the comparison of one‘s own income with that of other workers. The reference-point-dependent utility function tries to reflect the observation that many normal, healthy, and socially intelligent people do in fact care about their own payoffs relative to others. For some workers, it may be worthwhile to trade off a few thousand dollars of their own salary for a work environment where the relative pay structure is more to their liking (e.g., a feeling of relative high status in the $90,000 job being subjectively worth more than the extra $5,000 of income at the $95,000 job). It should be mentioned, however, that reference-point-dependent theories in behavioral economics are not entirely new. One finds interpersonal comparisons in Veblen‘s concept of conspicuous consumption from his classic The Theory of the Working Class (1899) and even earlier among some classical economists. Gigerenzer (2008) and Jorland (1987) have further questioned behavioral economists‘ historical reading of Kahneman and Tversky‘s (1979) prospect theory and its reliance on a reference point, which Gigerenzer and Jorland argue was already present in Daniel Bernoulli‘s original description of the expected utility function. According to Gigerenzer and Jorland, Bernoulli specified the argument of the expected utility function to be a change in, rather than absolute level of, wealth. This would imply that the reference point introduced by Kahneman and Tversky‘s prospect theory was in fact a re-introduction, interesting especially in light of Bernoulli and then Kahneman and Tversky‘s respective roles in the so-called repair program that began with expected value maximization (Güth, 2008; Gigerenzer, 2008, p. 90). Expected value maximization was, before Daniel Bernoulli, the unquestioned standard of rationality in gambling and games of chance. The St. Petersburg Paradox revealed the shortcomings of expected value

When trying to convince neoclassical economists who are skeptical about the need for behavioral economics, behavioral economists point to the improved ability of their psychology- inspired models to fit data collected from a variety of sources, including experimental, macroeconomic, and financial market data. Skeptics from outside behavioral economics have questioned whether the deviations from neoclassical assumptions have any important consequences for the economy as a whole, suggesting that they might perhaps ―average out‖ in the aggregate. Skepticism about the relevance of experimental data remains strong, with many doubts expressed about whether the college students who participate in economic experiments can be relied upon to teach us anything new about economics, and whether anything learned in one laboratory experiment can be generalized to broader populations in the economy—the so- called problem of external validity. Experimentalists have responded that the reason they carefully incentivize decisions by making subject payments dependent on their decisions is to make it costly for them to misrepresent their true preferences. Experimentalists have addressed the issue of external validity by going into the field with so-called field experiments, and by conducting experiments among different subpopulations, such as financial market traders, Japanese fishermen, and other groups of adult workers (e.g., Carpenter and Seki, 2006). Within behavioral economics, a different debate takes place. Among behavioral economists, despite a shared commitment to borrowing from psychology and other disciplines, there remains tension over how far to move away from constrained optimization as the singular organizing framework of neoclassical theory and in much of behavioral economics, too. An alternative approach, advocated by a minority of more psychology- and less economics-inspired behavioral economists, seeks to break more substantially with neoclassical economics, dispensing with optimization theory as a necessary step in deriving equations that describe

behavior. Constrained optimization, whether in behavioral or neoclassical economics, assumes that decision makers see a well-defined choice set; exhaustively scan this set, plugging each possible action into a scalar-valued objective function, which might include parameters intended to capture psychological phenomena; weigh the costs and benefits associated with each action, which includes psychic costs and benefits; and finally choose the element in the choice set with the highest value according to the objective function. There is very little direct evidence of people making decisions—especially high stakes decisions, such as choosing a career, buying a house, or choosing whom to marry—according to the constrained optimization process just described. In many real-world decisions such as those just mentioned, the choice set is impossibly large to clearly define and exhaustively search through. In other settings such as choosing a life partner or whom to marry, constrained optimization would be seen by some to violate important social norms. Instead, critics such as Gigerenzer and Selten (2001) attempt to base theory directly on empirical description of actual decision processes. Like other economists, these critics use equations to describe behavior. However, their behavioral equations skip the step of deriving behavioral equations as solutions to constrained optimization problems. To these researchers, theorizing and observing how decision makers deal with the overwhelmingly high-dimensional choice sets they face, quickly searching for a good-enough action and discarding the rest, is a fundamental scientific question of primary importance. Herbert Simon referred to such threshold-seeking behavior as satisficing as distinct from optimizing. Indeed, some leading voices in the applied area of marketing have recently discovered that they can make improved predictions about the way customers search for information and make purchase decisions by abandoning the optimization model. One reason why optimization

German mathematician, Ernst Zermelo, proved that there exists an optimal strategy in chess. But the combinatorics of many possible paths of play in a game of typical duration lead to a strategy space for each player that contains more elements than there are atoms in the universe. Although we know it exists, no computer can possibly compute the optimal chess strategy. Therefore, when describing the actual behavior of real-world chess champions (or expert computer programs that play chess), the debate in behavioral economics amounts to the following. Is the interesting scientific question whether chess champions fail to optimize (even when they win)? Or is the relevant scientific challenge to describe the thought processes that enable people to win? A more general statement of this problem of scientific relevance is whether to emphasize that human behavior differs from the prescriptions given by solutions to constrained optimization problems—or whether to focus on describing real human decision processes together with factors in the environment that enable people to succeed or fail. Many heuristics that are widely regarded as sub-optimal in behavioral economics could very well enjoy an alternative interpretation as brilliantly successful tools for making fast decisions in high-dimensional environments. As it stands now, much of the behavioral economics literature focuses on documenting deviations and biases with respect to the optimization model, which is equivalent to simply reporting that even grand master chess champions fail to play the optimal strategy in chess. An alternative empirical approach based on heuristics would study grand masters and identify the rules of thumb that they use to play the game as effectively as they do. The growing prominence of scholars and publications addressing these issues reveals the ongoing importance of debates over methodology and economic history (e.g., Gigerenzer, Todd and the ABC Group, 1999; Gilboa, Postlewaite and Schmeidler, 2004; Starmer, 2004, 2005; Heifetz, Shannon and

Spiegel, 2007; Bruni and Sugden, 2007; Caplin and Schotter, 2008; Hertwig and Hoffrage, forthcoming). Critics of the universal constrained optimization model that dominates in neoclassical and behavioral economics contend that a huge—and unrealistic—step of simplification is typically required when setting up the optimization problem: summarizing everything a person cares about with a utility function, or everything that the managers of a firm base their decisions on with a simple profit function. The methodological debate then becomes whether decision scientists learn more by assuming decision makers optimize in a pre-simplified world, or whether optimization can be productively abandoned in favor of direct empirical description of the simplifying strategies people actually use in real-world environments that are typically many times more complex than the game of chess.

Methodological Pluralism Another theme in behavioral economics derives from its willingness to borrow from psychology and other disciplines such as sociology, biology and neuroscience. To appreciate why methodological pluralism is characteristic of behavioral economics, one should recall that in neoclassical economics there is a singular behavioral model applied to all problems as well as a number of prominent efforts in economic history to expunge influence from other social sciences such as psychology and sociology. Although the structure of choice sets and the objective functions change depending on the application, contemporary economists typically apply the maximization principle to virtually every decision problem they consider. Consumer choice is modeled as utility maximization; firm behavior is modeled as profit maximization; and the evaluation of public policy is analyzed via a social welfare function whose maximized value

Another, albeit more minor, point that occasionally gives rise to confusion is the multiple uses of ―psychology and economics‖ as a descriptor of an academic subfield, which can indicate subtly different communities of researchers depending on whether it is regarded as a subfield of economics or of psychology. The subfield of psychology referred to as ―psychology and economics‖ overlaps in terms of subject matter and actively contributing scholars. Behavioral economists are sometimes regarded by psychologists, however, to overstate the extent to which the work of behavioral economists is actually informed by the broad and heterogeneous research programs within psychology.

Behavioral Economics and Experimental Economics Strong connections between behavioral and experimental economics can be seen in behavioral economists‘ reliance on experimental data to test assumptions and motivate new theoretical models. There nevertheless remains a distinction to be made (Camerer and Loewenstein, 2004). Some experimental economists do not identify with behavioral economics at all, but rather place their work firmly within the rational choice category, studying, for example, the performance of different market institutions and factors that enhance the predictions of neoclassical theory. Experimental economics is defined by the method of experimentation whereas behavioral economics is methodologically eclectic. The two subfields have subtly different standards about proper technique for conducting lab experiments and very different interests about the kinds of data that are most interesting to collect. Therefore, it is incorrect to automatically place experimental work under the heading of behavioral economics. In the other direction, there are many behavioral economists working on theoretical problems or using non-experimental data. Thus, although behavioral and experimental economists frequently

work complementarily on related sets of issues, there are strong networks of researchers working in the disjoint subsets of these subfields as well.

Frequently Discussed Violations of Internally Consistent Logic

This section describes several well-known violations of the rational choice model based on reasoning that allegedly suffers from internal inconsistency. The following example about deciding where to buy a textbook illustrates the kind of inconsistencies that are frequently studied in behavioral economics. Readers are encouraged to decide for themselves how reasonable or unreasonable these inconsistencies in fact are. Suppose you are shopping for a required textbook. A bookstore across the street from where you work sells the book for $80. Another bookstore, which is 15 minutes away by car or public transportation, sells the book for only $45. Which do you choose: A) Buy the book at the nearby store for $80, or B) Buy the book at the farther-away store for $45? Just as standard theory does not prescribe whether it is better to spend your money buying apples versus oranges, so, too, standard economic theory takes no stand on which choice of stores is correct or rational. But now consider a second choice problem. Suppose you are buying a plane ticket to Europe. The travel agent across the street from where you work sells the ticket for $1,120. Another travel agency, which is 15 minutes away by car or public transportation, sells the same ticket for $1,085. Which do you choose: C) Buy the ticket from the nearby agency for $1,120, or D) Buy the ticket at the farther-away store for $1,085? Considered in isolation, either A or B is consistent with rationality in the first choice problem, and either C or D can be rationalized in the second choice problem—as long as these

Skepticism over the importance of such violations of axiomatic rationality is discussed in a subsequent section under the heading Rationality.

Endowment Effect Suppose you walk into a music store looking for a guitar. The very cheap ones do not produce a sound you like. And most of the guitars with beautiful sounds are priced thousands outside your budget. You finally find one that has a nice sound and a moderate price of $800. Given the guitar‘s qualities and its price, you are almost indifferent between owning the guitar and parting with $800, on the one hand, versus not owning it and hanging on to your money on the other. You go ahead and buy the guitar. After bringing it home, enjoying playing it, and generally feeling satisfied with your purchase, you receive a phone call the very next day from the music store asking if you would sell the guitar back. The store offers $1,000, giving you an extra $200 for your trouble. Would you sell it back? According to the standard cost-benefit theory, if you were indifferent between the guitar and $800, then you should be more than happy to sell it back for anything over $800—as long as the amount extra includes enough to compensate for the hassle, time and transport costs of returning it to the store (and also assuming you haven‘t run into someone else who wants to buy the guitar and is willing to pay a higher price). Hoping to bargain for a higher offer from the music store, you might demand something far above $800 at first. But after bargaining, when facing a credible take-it-or-leave-it last offer, anything that gives you $800 plus compensation for returning to the store should leave you better off than holding onto the guitar. Based on data showing the prevalence of the endowment effect, however, behavioral economists would predict that you probably will choose to hang onto the guitar even if the guitar

store‘s offer climbed well over $1,000. The endowment effect occurs whenever owning something shifts the price at which one is willing to sell it upward to a significantly higher level than the price at which the same person is willing to buy it. In the neoclassical theory taught in undergraduate textbooks with demand curves and indifference curves, an important maintained assumption that is not frequently discussed in much depth is that, for small changes in a consumer‘s consumption bundle, the amount of money needed to just compensate for a reduction in consumption is exactly equal to the consumer‘s willingness to pay to acquire that same change in consumption. This equivalence between willingness to accept money for a reduction in consumption and willingness to pay money to buy the amount by which consumption was reduced implies reversibility of demand and indifference curves, which the endowment effect opens up to question. Another description of this assumption is that there is a rock-solid stable relationship between quantities of consumption and their subjective valuations that does not change depending on what a person is currently consuming. When current consumption affects how I value all other possible combinations of consumption, then all bets are off—demand curves do not exist. Or to put it more precisely, a single person has many different demand curves, one for each point in the consumption space at which he or she consumes. Kahneman, Knetsch and Thaler (1991) presented data showing that randomly assigned ownership systematically increases subjective valuations of goods and services. In one famous experiment, coffee mugs were randomly distributed to half the experimental subjects. Those who did not receive mugs were asked to submit bids to buy a mug, and those who owned a mug submitted offers to sell their mug. There was no negotiating or haggling. The experimenter collected bids and offers, and used these to find a market price at which the quantity supplied