A rule of one.

A rule of one--one rational person is enough to make a market; one irrational person is enough to break a game. Let me explain. The power of the market as an exchange institution rests in its ability to move resources from low- to high-valued uses. The market works if one person is rational enough

to look for arbitrage opportunities among the other irrational players, he buys low and sells high, the market collects and disseminates this information such that others learn or exit. In the limit, this person works the collective to exhaust all gains from trade, as predicted (Hayek 1945). Evidence from four decades of economic laboratory experiments has established how institutions like the double auction market can capture nearly all potential gains from trade in a variety of situations with an assortment of inexperience and experienced traders (Smith 2003). People in this world can be considered econobots, in that they act like automatons who neither set prices nor need to account for others' attempts to affect the market (Gode and Sunder 1993).

In contrast, it suffices for one person to behave irrationally for a game to yield outcomes that deviate from the predicted actions of rational behavior. Irrationality means he takes incredible actions or holds unreasonable beliefs about the other players' skill. Behavior differs because the other players must drop their level of play to that of the least rational player, i.e., the weakest link drives the game. If an irrational player does not know how to play the game because he does not know how to use backward induction or does not think strategically, the other players have to adjust their play accordingly. Now people do not play the game as predicted because rational players cannot arbitrage the irrational players. The irrational player himself has to recognize the disadvantages, if any, from playing their incredible actions or holding their unreasonable beliefs. And in some games, playing irrationally pays off when it avoids low-payoff equilibria like those that exist in the one-shot prisoners' dilemma or in the centipede games (Kreps et al. 1982; McKelvery and Palfrey 1992). The poor correspondence between rational choice theory and observed behavior in gaming experiments has lead some to argue for a "behavioral game theory" (e.g., Camerer 2003). Behavioral game theory asks researchers to combine theoretical equilibrium analysis with more work based on empirical pattern recognition.

Many of us, however, lean on rational choice theory as our tool to guide resource management and public policy challenges like environmental protection or education quality. To some, it makes more sense if people make, or act as if they make, consistent and systematic choices toward certain and risky events. In mainstream economics, the concept of rational behavior is defined as a social construct--an economic choice within the social institution of mercantile exchange; not an isolated decision made in a private bubble (Arrow 1987). Markets create rationality in the population by putting a cost on irrational behavior. Evidence from the lab shows how people in market-like settings act more rational because rationality now pays; whereas those outside markets can afford to be irrational because others do not exploit their choices (see, e.g., Berg, Dickhaut, and O'Brien 1985; Cox and Grether 1996; Loomes, Starmer, and Sugden 2003). Anomalous behavior tends to disappear with arbitrage once a person decides he prefers more money to less.

Economics walks a razor's edge--rational choice theory does a reasonable job in understanding market behavior of a de facto econobot, but the method is more problematic when trying to predict the gaming behavior of a minor-league Machiavellian (figure 1) making nonmarket allocation decisions like those associated with environmental protection. The edge of rational choice becomes even thinner when trying to explain or predict the behavior of an isolated individual contemplating a nonmarket decision like life or death, i.e., Hamlet pondering outside the socializing influence of his confidants. Unless we think market-like institutions are so powerful that they work to condition people into making more rational choices for nonmarket decisions, we should be careful to accept uncritically the proposition that a person filling out a mail survey in the privacy of his own living room would rationally state a hypothetical monetary value to reduce risks to life or limb (Knetsch 1989).

[FIGURE 1 OMITTED]

This challenge raises two sets of nagging questions. The first set raises the possibility we need to make a transition from our current neo-classical framework of context-independent preferences/choice to the behavioral structure of context-dependent preferences and choice. How rational should people be for policy? Does the context of choice affect preferences? Do incentives matter in the way we think they do? What role do emotions play in economic behavior? How complicated is a utility function? How do institutions affect preferences?

This set of questions led me to a crossroads in my research aimed at understanding the gap between observed and predicted behavior. Should one try to explain the behavioral gap ex post through advances in behavioral economics or try to eliminate the gap ex ante through institutional design? Explaining the gap ex post requires inundating oneself into the ever-expanding taxonomy of irrational or anomalous behavior (e.g., endowment effect, loss aversion, preference reversals) based on the pioneering work of Kahneman and Tversky (2000). Accounting for the gap leads one's research into either inventing new degrees of behavioral freedom based on emotions, information processing, the formation of perceptions, or the processes used when thinking about things (McFadden 1999). The crux of this research rests in adding emotional degrees of freedom into a utility function (e.g., regret, suspicion, envy), or probability weight schemes around how people's beliefs about states of nature. This research has also lead scholars to introduce a New Paternalism aimed at helping people who know better help themselves overcome self-control, i.e., libertarian paternalism to help people save more, exercise more, eat less, smoke less, and so on (Thaler and Sunstein 2003; Hsee and Hastie 2006).

My decision was to take the second path--try to eliminate the gap ex ante through the design of institutions. Without psychoanalyzing my decision, it makes more sense to work with something economists are generally more comfortable with than emotions--institutional design. The goal is to construct an institution or context that can provide the incentives to induce people to act more rational (whether they actually are or not is another question). If one is successful at institutional design, two useful points emerge. First, rational choice theory is supported under these institutional conditions, which might be relatively robust and which might parallel real world situations (see Henrich et al. 2005, and the lively rebuttals). This keeps our neoclassical modeling efforts developed over the last few decades intact; no reason to add new emotions or weighting schemes. Second, if such institutions can be designed, one might argue a breakdown in rational choice is not behavioral failure. Rather one might make the case these anomalies are a new form of market failure: the failure of the exchange institution to provide a money pump or arbitrage mechanism (buy it low, sell it high) sufficiently strong to induce rational choice. Again we are familiar with contemplating traditional market failure (e.g., externalities, public goods), and we can avoid the push for the new paternalism based on behavioral failure.

The set of questions that arise along this path include what institutions produce the behavior economists assume? Can we create new institutions that better spotlight opportunity costs of irrational behavior? Do these institutions exist in the wilds? Are they reasonable and robust for market goods? Could they be reasonable and robust for nonmarket goods? This paper documents some of our recent attempts to address aspects of these broad questions in the business of institutional ex ante anomaly removal. We consider three specific examples of behavior in the lab given alternative institutional structures and incentives; some work better than others at inducing rational behavior within market and nonmarket allocations. Before going through the examples, first consider the common core that has served as the analytical foundation underlying my behavioral work over the last two decades.

Common Core

The behavioral underpinnings of my institutional design research comes from a common core--the idea that people make choices within institutional contexts to increase the probability that good things happen and bad things do not. In brief, the theory of endogenous risk reflects the view that a person has preferences over outcomes and preferences over the lotteries that define those outcomes, and makes private and collective choices to secure those more preferred lotteries. These preferences for lotteries are conditioned on the institutional setting. People self-protect in many ways by expending resources on private risk reduction actions and institutional design to give people incentive to protect themselves (e.g., water filters, preventive care, and nonlinear payoffs). The framework captures the idea that people affect nature, nature affects people, and this nature via nurture interaction is heightened or hindered by the institutions we create (Shogren and Crocker 1991).

To make this more concrete, suppose a risk-averse person decides ex ante how to invest resources to control privately the risks he confronts, i.e., loss of some valuable personal asset such as his house or his health harmed by an environmental hazard. Because of moral hazard, adverse selection, and nonindependence of risks, this person cannot acquire enough market insurance to assure his ex ante utility level is maintained whether or not the harm occurs. Assume two states of the world, good and bad, for simplicity such that expected utility is presented by:

(1) EU= p(z;[bar.p]|I)U(w - L(x) - z - x|I) + (1 - p(z;[bar.p]|I))U(w - z - x|I)

where U(w - L(x) - z - x|I) and U(w - z - x|I) represent the utility of a bad and good state conditional on the institutional circumstances, I; p(z; [bar.p]|I) and (1 - p(z; [bar.p]|I)) represent the probability a person realizes the bad state and suffers a monetary loss, L, or not, again conditional on the institutional design; z is self-protection, x is self-insurance, which represents the nurture side of the nature via nurture worldview. Let p represent an exogenous, uncontrollable probability of a bad outcome, i.e., background risk provided by nature. Self-protection represents private investments to increase the probability that a good state occurs; self-insurance are expenditures to reduce the severity of the bad state if it is realized. Assume self-protection reduces the probability, [partial derivative]p(z; [bar.p]|I)/[partial derivative]z < 0, and greater exogenous probability increases it, Op (z;[bar.p]|I)/ [partial derivative][bar.p] > 0. Losses are a function of investments in self-insurance, x, such that L' < 0, where prime is the derivative. The price of self-protection and self-insurance are normalized to unity for simplicity.

This represents the rational choice benchmark of a person who invests resources to try and control the events that surround him or her. If this person is subject to behavioral failures/imperfections, one needs to add degrees of freedom to the model such that departures can be rationalized ex post by the new auxiliary assumptions. In particular, we add an innate emotional vector, [theta] = ([[theta].sub.1], [[theta].sub.2], ... [[theta].sub.k]), of k different emotional attributes (altruism, social preferences, and so on) into the utility function, [[theta].sup.+] for gains and [[theta].sup.-] for losses from the status quo. We also add the probability weight [pi] to capture the prospect theory notion that people alter their subjective probabilities based on behavioral regularities (e.g., overweigh low probability chances), such that [pi](p(z; [bar.p]|I)) and (1 - [pi](p(z;[bar.p]|I))).

The goal of ex ante institutional design is to find or design an institution, I, which is powerful and robust enough to remove all the nonmarket emotional, [theta], and probability weights, [pi], from the decision making process. This is the money pump argument--a strong institution to discipline inconsistent choices can produce more consistent choices (Nau and McCardle 1991; but see Cubitt and Sugden 2001). Now these behavioral traits are too costly to entertain in one's decision making for some market or nonmarket good. Implicit in this view is that these emotions and weights are not primitive absolutes, or at least are not so primitive and buried in the genetic code that one can never learn to "control one's emotions" and "control expectations" relative to, say when the person was a child. My claim is not that these natural primitives disappear from a person's core personality. Rather my point is that these primal instincts are relatively less influential on choice relative to their innate and instinctual preferences under the right institutional circumstances (see Pinker 2003 or Lynne 2006 for another view).

We use the experimental method to explore this question of institution-induced rationality. Since the work of Bohm 1972, Menkhaus et al. 1992, and others, it is becoming commonplace to use experiments in resource and agricultural economics (e.g., see Corrigan and Rousu 2006). Implicitly experimentalists accept the point Fredrick Waugh made in a letter to Ezekiel: "I want to get hold of anything new and worthwhile ... " (Bessler and Dearmont 1996). The experimental method has proven itself useful to explore questions of institutional impacts on behavior, and beyond (Plott 1994).

Lessons from the Lab

Consider three lessons learned from our recent experimental work on how institutional context matters, both for more or less economic efficiency, when thinking about policy questions within the rational choice framework. The goal of walking you through these examples is to illustrate how one's choice of institutional context, I, affects our ability to predict behavior relative to the rational choice benchmark. The selection of the institution can either accentuate or attenuate the impact of our nonmonetary emotional factors, [theta], on our decisions-the impact, however, is not always predictable ex ante.

    Lesson #1: We can induce more rational behavior
    by creating an institution with real
    market-like arbitrage that disciplines irrational
    behavior. People can generalize and
    transfer this learned behavior to similar nonmarket
    choices without arbitrage and to
    distinctly different choices, even if arbitrage is
    hypothetical People act more rational by revising
    their stated monetary values rather than
    by changing preferences, i.e., preferences can
    be fixed, not fangible.

Evidence

Research has shown how people acting outside the realm of market-like discipline exhibit behavior inconsistent with rational choice theory (Baron 2000). Classic examples include the A1lais paradox, the Ellsberg paradox, status quo bias, anchoring, framing, endowment effects, loss aversion, overweighting low-probability events, and preference reversals among others (McFadden 1999). People are unduly influenced by how a lottery is framed. This violates a basic principle of rational choice, which says that preferences should be invariant to how gambles are described. But more recent lab evidence suggests some of these individual lapses in rationality can be corrected by the social discipline provided by an active exchange institution like a market. An institution with market-like arbitrage can induce seemingly irrational people to act like they are more rational once they experience the opportunity costs of their irrational behavior. Once a person decides he does not like being turned into money pump, he stops making inconsistent choices.

Tversky and Kahneman (2000, p. 223) take issue with this perspective: "[t]he claim that the market can be trusted to correct the effect of individual irrationalities cannot be made without supporting evidence, and the burden of specifying a plausible corrective mechanism should rest on those who make the claim." Consider some evidence for the case of preference reversals. Recall a person reverses his or her preferences when he faces two lotteries of equal expected value but different risk, and he indicates a choice preference for the safer lottery but states a value preference for the riskier bet (Lichtenstein and Slovic 1971). But results from experiments are that once this isolated person makes his choices within the social context of others who could exploit his or her inconsistent decisions, he stops reversing his preferences. Chu and Chu (1990) find that one or two experiences through a money pump are enough to get most people to act rational.

In addition, the power of arbitrage can extend beyond the reach of the market-like institution to nonmarket choices. Over the last decade I have stressed this point many times in many outlets, but it is worth repeating again. In Cherry, Crocker, and Shogren (2003), we illustrate how this rationality spillover can occur in an experimental design that creates choices within both a market and nonmarket setting. The market uses an arbitraged system design to create a money pump for anyone who reversed his preferences, whereas the nonmarket setting did not. With arbitrage, rents from subjects who reversed preferences were extracted in three steps. The market sold the least preferred/most valued lottery to the subject; next it traded the most preferred lottery for the least preferred lottery; and finally, it bought the most preferred/least valued lottery from the subject. The subject was left with only a hole in his pocket.

Without arbitrage, we observed people reversing preferences about 30%-40% of the time. Once arbitrage is introduced, preference reversals decline significantly, as expected. We also found that this learned rationality could spill over to their choices in the nonmarket setting. People stopped reversing their preferences in the nonmarket setting even though (a) no direct arbitrage took place, (b) the nonmarket choices were hypothetical, and (c) the lotteries switched to wildlife experiences in Yellowstone National Park.

Intense market experience was a powerful factor on otherwise anomalous choices and statements of value. A critic might call this brainwashing or a "harsh penalty" solution that reveals the economist's "defensive posture" about the reversal phenomena (Slovic 2000, pp. 493-4). A more constructive response is to explore how robust this rationality spillover effect could be as we lightened up on the intensity of the arbitrage mechanism. In Cherry and Shogren (2006), we considered whether a one-time experience with a real or hypothetical arbitrage mechanism would work, and whether it would actually create a rationality crossover--people become more rational in related but distinct decision tasks. We designed a three-stage experiment that elicits choices and stated values, both with and without the experience of arbitrage. We found that rationality crossovers also exist, in which people change behavior across contexts. Irrationalities were not eliminated, but occurred substantially less frequently given a one-time experience with arbitrage. Stated values for safer food fell by 30%-40% and the frequency of inconsistent risk preferences was cut in half. In addition, we found these results hold even under a very weak type of arbitrage-cheap talk, a nonbinding explanation on how arbitrage works and its consequences on one's wallet.

If markets can help induce more rational behavior, the open question is why? Are a person's preferences for risky events fixed or fungible given market experience? Most economists see fixed preferences as a valuable precept in rational choice theory; context-independent preferences are a fundamental building block in demand theory which has served them well in describing behavior within active exchange institutions. Psychologists counter that preferences are fungible and context-dependent. Noneconomic contextual cues affect preferences more than most economists will admitt (Tversky and Simonson 1993; Ariely, Loewenstein, and Prelec 2003).

Knowing whether preferences are "transient artifacts" contingent on context or not matters because if they are fungible, then so are the welfare measures used in cost-benefit analyses to rationalize or reject regulations to protect health and safety. In Gunnarsson, Cherry, and Shogren (2003), we estimated an empirical model of the preferences for risk and skewness, the love of the long shot, in market-like and nonmarket settings. The results show that preferences remained stable even as arbitrage removed preference reversals. People stopped reversing preferences with arbitrage not because their preferences were fungible, but because they initially overpriced the risky long shot. Arbitrage caused people to reconsider and correct the inconsistency of their preferences and values, but the reconciliation of preferences and values arose from value adjustments not preference adjustments.

In summary, our findings do not prove that arbitrage made people think more rationally; rather our results show that people can act as if they were more rational, whatever the psychological motivation. The institutional structure created by the money pump led to the behavior predicted by rational choice theory implied for endogenous risk in expression (1).

    Lesson #2: We can induce more rational behavior
    by making the institutional context
    more complex rather than less complex. Evidence
    suggests a less complex institutional context
    does not necessarily lead to more rational
    gaming behavior. We find this result holds.for
    equilibrium behavior in coordination games
    and for bidding behavior in Vickrey (1961)
    second-price auctions.

Evidence

First, consider behavior within a classic coordination game with differing degrees of complexity. Coordination games represent those circumstances when people need to coordinate over multiple equilibria, which can usually be Pareto ranked (Schelling 1960). Coordination games are everywhere in economics and society, e.g., macroeconomic policy, agglomeration decisions, network externalities, team production, and imperfect competition. Coordination games tested in the lab use normal form games that explicitly reveal the entire strategy set for two players and the payoff matrix for each combination. Evidence suggests people coordinate to the payoff dominant outcome using a normal form game more frequently relative to a risk dominant outcome if they can communicate, have long time horizons, and have access to markets selling rights to play.

Figure 2 illustrates an 8 x 8 symmetric normal form game, in which two players, A and B, have 8 actions each (1 - 8), creating 64 payoff cells. From a behavioral perspective, a normal form game boils the coordination problem to its essentials--actions and payoffs. A normal form is an institution that reduces a player's decision costs by providing complete and perfect information about strategy sets and payoffs for every possible combination of actions. In this coordination game, there are four pure strategy Nash equilibria--actions (4,4), (5,5), (7,7), and (8,8). Each of these four action-action combinations is a Nash equilibrium because once that action is selected by both players neither player has a unilateral incentive to change his action. Action 5 is defined as the risk dominant strategy for each player--if he selects action 5, he earns 111 regardless of what the other player does, i.e., actions (5,5) is the risk dominant Nash equilibrium. Actions (4,4) represent the payoff dominant Nash equilibrium since both players earn the most if they coordinate to this action. Actions (7,7) and (8,8) represent intermediate payoff Nash equilibria. Both the payoff dominant and risk dominant (safe bet) strategies are natural focal points within the game.

In Parkhurst and Shogren (2005), pairs of subjects played this game over twenty rounds. Economic efficiency is maximized within this game when a pair of subjects selects the payoff dominant Nash equilibrium, (5,5). Each subject (a) could send one pre-play nonbinding message per round to the other player about what action to take, (b) selected one action after sending and receiving the messages, (c) selected his action, (d) learned the action of the other player, and (e) determined his payoff for that round. Table 1 shows that individuals played the payoff dominant strategy about 82% of the time in rounds 1-5, which increased to nearly 94% by rounds 16-20. People had a relatively easy time coordinating to the private and social optimum of this game.

This game was created to capture the coordination decisions of landowners offered a subsidy scheme to retire voluntarily their land into a contiguous habitat reserve to protect, say an endangered species. The normal form game is highly abstract, stripping land retirement decisions down to actions and payoffs. There is no explicit spatial dimension, actions are limited to eight, payoffs to sixty-four, and all information is complete and perfect. This game is a serious abstract from reality, which troubled many colleagues when we discussed the implications for policy. The question we always received was how robust is coordination when we move away from the traditional normal form setting toward a more realistic setting.

Figure 3 illustrates the grid game--our attempt to move the land retirement coordination problem closer to reality by adding an explicit spatial dimension. We increase the institutional complexity by examining the behavior of four landowners who own 25 units of land each, which are valued at a $2, 4, 6, 8, or 10 per unit. The conservation target is to get each landowner to retire 5 units such that it creates one long 20-unit corridor on the most expensive land ($10), say prime agricultural land along a river. If you were playing, your land would be the 25 units in the northwest; and you would have three neighbors, south, east, and southeast.

[FIGURE 3 OMITTED]

More complexity is added in that the land retirement subsidy is no longer embedded into the payoffs as in the normal form game; rather the subsidy used one flat fee and two spatial agglomeration bonuses: (i) a per land unit benefit, $3, a flat fee per retired land; (ii) an own shared border benefit, $8, an extra payment for every border shared between two of his own retired land units; and (iii) a row shared border benefit, $16, a payment for every border shared between two retired--one own retired unit and a retired unit from the southern neighbor. Given this complexity, all experiments were run on a computer network. Each player had a grid calculator in which he could play around with different land retirement combinations--his own and neighbors--and see how different actions affected his payoffs.

In this setting, neither the Pareto nor risk dominant strategies are spotlighted as in the normal form game. Each player now has 68,000+ choices (rather than 8 as in the normal form), with over (68,000) (4) potential outcomes (rather than 64). This created over 9,400 Nash equilibria (rather than 4) in this grid game, in which only the targeted long corridor is the payoff dominant equilibrium. Again players could send one nonbinding message per round to other players in the group.

We have seriously increased the complexity of the coordination problem--2 to 4 players, 8 to 64,000+ actions, 64 to (68,000) (4) potential outcomes, complete to incomplete information on the payoffs. Table 1, however, shows that people still coordinated successfully within this complex environment. The number of players selecting the payoff dominant strategy increase from less than 40% in rounds 1-5 to 100% in rounds 11-20. We cannot reject the null hypothesis--people in the grid game were equally likely to coordinate to the dominant Nash equilibrium as in the simpler normal form game. Although players performed better in the early rounds of the normal form game, subjects in the grid game perform at least as well in rounds 6-20. The institutional structure led to the behavior predicted by rational game theory.

Consider now another example of increased complexity within a game--here we consider bidding behavior in a Vickrey (1961) second price auction given two distinct institutional contexts. The second price auction is popular with researchers because it is weakly demand revealing--he should bid his true value irrespective of what other players do in the auction. In addition, the market-clearing price is endogenous and the allocation and pricing rules are easy to explain to people. But whether bidders understand these rules in practice is another question. Evidence from induced value studies suggests that bidders do not always bid their true value, especially bidders who are unlikely to be in the hunt for the good, i.e., values well below the potential market-clearing price. Insincere bidding occurs when people either do not understand the incentives or the incentives are not enough to get them to pay full attention to the game.

In Shogren, Parkhurst, and McIntosh (2006), we explore whether recasting the auction as a tournament with a nonlinear payoffs could get people to bid their true values more frequently than the traditional method. Tournaments exist in the real world to get people to pay attention to small differences in measurable performance (Ehrenberg and Bognanno 1990). The question is whether we can induce the bidding behavior theory predicts using an institutional design that does not lend itself to precise theoretical predictions.

The standard and tournament auction designs were identical except for the payoff scheme. In the standard auction, take-home pay was: $5 + $5 * (total points earned over all trials for winning the auction); this is the typical design, in which a player could earn about $20 on average. In the tournament, take-home pay depended on points earned by the person relative to points earned by all other participants over all trials. Tournament payments were: $120 to the person with most total points: $80 to the runner up; $50 for third; $30 for fourth; $15 for fifth through seventh; and $5 for eighth through tenth. For each round, induced values were randomly drawn from a uniform distribution of [$0.00, $20.00] in ten-cent increments. Each auction had ten bidders; two sessions were run for each treatment for a total of forty bidders.

We tested for truthful bidding behavior using conditional panel regression methods and found the unexpected: we reject the null hypothesis of truthful bidding for the standard auction, but we do not reject it for the seemingly more complex tournament setting. The more complex tournament institution generated more rational bidding behavior than the standard auction. This unanticipated finding is consistent with three behavioral stories. First, the tournament payoffs were framed such that winning points given the nonlinear payoffs made points appear more valuable in the tournament. Alternatively, since the auction tournament was a repeated game of incomplete information with almost common knowledge, a bidder works backwards eliminating all dominated strategies, which leads him to the weakly dominant strategy of bidding his true value in each round. More likely, however, is the third explanation based on Heiner's (1983) theory on the origins of predictable behavior. Heiner (1983) posits the more uncertainty, the more predictable a person's behavior. When you have no clue what to do, you do what you know based on the information available. The only information each player had in a bidding round was his true value (and own accumulated points). He had no legitimate chance to strategize against an opponent. These results do not contradict the idea that the "predictable" behavioral rule is to bid one's induced value.

In summary, we find more rational behavior in more complex settings, both for a coordination game with over 9,000 Nash equilibria and second price auction tournament with incomplete information. Perhaps models like Heiner's (1983) predictability theory given alternative institutional contexts deserve a closer examination when considering how environmental and agricultural policies create complex circumstances that do not lend themselves to oversimplification. If so, this suggests it is worthy to devote more attention to institutional design relative to ex post behavioral rationalizations.

    Lesson #3: The simplest institutional frames
    affect the simplest games in common senses
    ways not predicted by standard rational choice
    theory. We can induce more rational behavior
    by creating an institutional context that allows
    people to be less sympathetic to others.

Evidence

We now consider how contextual institutional changes, simple, subtle, or otherwise, can affect rational behavior. Over the last decade, economists have tried to isolate and measure social preferences, i.e., how people treat other people, in a monetary fashion. Such "other-regarding behavior" is frequently observed in bargaining games. Consider, for example, the Anonymous Dictator game (see Hoffman, McCabe, and Smith 1996). This game controls self-interested strategic behavior by giving a person complete control over the distribution of wealth, and complete anonymity from all others including the experimenter. While theory predicts people with complete control and complete anonymity will offer up zero to others, evidence reveals they frequently share some wealth nearly half of the time. Social preferences are another example in which a person's behavior differs from that predicted by game theory, and supports the call for a new behavioral game theory.

In Cherry, Frykblom, and Shogren (2002), however, we created an institutional context in which 95% of all dictators follow game-theoretic predictions. We rejected social preferences as a context-free concept. Our simple change in experimental design had people bargain over earned wealth rather than unearned wealth. The experiment had two stages--earnings and bargaining. In the earnings stage, people earned money by taking a quiz using questions from the Graduate Management Admission Test (GMAT). Each person who answered 10+ questions correctly would earn $40 (high stakes); otherwise, one received $10 (low stakes). In the bargaining stage, people were randomly matched to form bargaining pairs. These dictators now were asked to dictate a split of his or her wealth with another person in another room. Administrators carried the offers from the dictator to the other person, and final earnings were paid according to the dictated split.

Based on the two-stage design, three treatments were run with high and low stakes session: baseline, earnings, and double blind with earnings. In the baseline treatment, people did not participate in an earnings stage prior to bargaining. As in previous studies, dictators in Room A were provided their wealth by the experimenter. Rather the monitor allocated a windfall of either $40 or $10. In the earnings treatment, dictators were told that the other person "has not had the opportunity to earn any money." The double blind with earnings treatment made the high and low stakes dictators anonymous to each other. Their results indicate that other-regarding behavior is greatly diminished when bargaining involves earned wealth, and this behavior is nearly eliminated when earned stakes are combined with anonymity. Dictators bargaining over earned wealth were more self-interested than observed in previous studies; and when they had complete anonymity, selfless behavior is essentially eliminated. This result suggests other-regarding behavior arises entirely from strategic concerns within the context that was provided.

Now consider how a simple change in institutional context affects dictator behavior. Figure 4 shows a 3 x 2 thought experiment conducted in my Global Economics Issues class last spring. The students were asked to do the following hypothetical experiment--think of yourself as a dictator and decide how much you will give to (out of $100) an unknown person in each of six contexts: either (i) earned or (ii) windfall wealth; and either (a) the other person has not had the opportunity to earn any money, (b) The other person has not earned any money, and (c) the other person has decided not to earn extra money. Regardless of the context all decisions should be the same if the person is rational as defined by sub-game perfection--he should give nothing to the other person in all cases. Figure 5 shows the results based on 88 subjects. We see that rational choice theory predicted behavior best in the context of {earned wealth/decided not to earn}--the average person kept nearly $99 for himself. Rational choice theory did the worst in predicting {windfall wealth/no opportunity to earn}--the average person kept about $70 for himself. Intuitively, most lay people think this makes perfect sense. You had the good luck to find $100 without laboring why not share some of it with others. Institutional context played a significant role in generating behavior implicit in rational game theory.

[FIGURE 5 OMITTED]

Concluding Remarks

A rule of one: one person is enough to make a market; one is enough to break a game. But other rules of one also exist. In the theatre, the rule of one means that an audience grants the storyteller one unexplained gap in the plot before they find the story incredulous. The story told herein has more than one hole in its plot. The examples presented in this paper raise more questions than answers: can we find a less brutal way to design institutions to generate rational behavior? Why does a more complex decision environment lead to behavior more predictable by rational choice theory? Who is a more insightful guide into the irrational elements of human decision making--William Shakespeare or a Nobel laureate economist? Do nonmarket money pumps exist unrecognized for nonmarket goods? If not, should we create them? Do we want rationality to transfer across market and nonmarket contexts? Under what institutional conditions does it work? If we solve a market failure problem without addressing potential behavioral failures, have we walked into a new version of the theory of second best? That is, when you fix only one of two problems, might you actually make things worse? Or is behavioral failure really just market failure in disguise?

Attempts to understand the nature of these questions has pushed my research into one of the oldest debates in science: nature versus nurture, which in my opinion is more accurately encapsulated by Ridley's (2003) turn of phrase nature via nurture (although see Pinker 2003). The decision into this old debate brings all our questions around more than full circle--to the idea of a paleoeconomics. By paleoeconomics, we mean the exploration of the prehistoric conditions under which human preferences were being formed and evolving based on both the harsh world they faced and the institutions they invented (e.g., Ofek 2001; Bowles, Choi, and Hopfensitz 2003; Horan, Bulte, and Shogren 2005). If behavioral anomalies and social preferences are hard wired, we want to know why and how. Basic economic forces of scarcity and relative costs and benefits have played integral roles in shaping societies throughout recorded human history. No reason exists today to discount either the presence or potential impact of economics in the pre-historic dawning of humanity, either on the co-evolution of institutions or the evolution of basic preferences for risk bearing and definitions of happiness. A better understanding of the behavioral underpinnings of choice within market and nonmarket contexts might have to take a step back in history to explore the co-evolution of nature and nurture via institutional choice.

References

Ariely, D., G. Loewenstein, and D. Prelec. 2003. "Coherent Arbitrariness: Stable Demand Curves without Stable Preferences." Quarterly Journal of Economics 118:73-105.

Arrow, K. 1987. "Rationality of Self and Others in an Economic System." In R. Hogarth and M. Reder, eds. Rational Choice: The Contrast between Economics and Psychology. Chicago: University of Chicago Press.

Baron, J. 2000. Thinking and Deciding, 3rd ed. Cambridge: Cambridge University Press.

Berg, J., J. Dickhaut, and J. O'Brien. 1985. "Preference Reversal and Arbitrage." In V. Smith, ed. Research in Experimental Economies: Vol. 3. Greenwich, CT/London: JAI Press.

Bessler, D., and D. Dearmont. 1996. "Ceteris Paribus: A Revolution with Agricultural Econometrics." European Review of Agricultural Economics 22:272-80.

Bohm, P. 1972. "Estimating Demand for Public Goods: An Experiment." European Economic Review 3:111-30.

Bowles, S., J.-K. Choi, and A. Hopfensitz. 2003. "The Co-evolution of Individual Behaviors and Social Institutions." Journal of Theoretical Biology 223:135-47.

Camerer, C. 2003. Behavioral Game Theorv. Princeton, N J: Princeton University Press. Cherry, T., T. Crocker, and J. Shogren. 2003 "Rationality Spillovers." Journal of Environmental Economics and Management 45:63-65.

Cherry, T., P. Frykblom, and J. Shogren. 2002. "Hardnose the Dictator." American Economic Review 92:1218-21.

Cherry, T., and J. Shogren. 2006. "Rationality Crossovers." Working paper, University of Wyoming.

Chu, Y.-E, and R.-L. Chu. 1990. "The Subsidence of Preference Reversals in Simplified and Marketlike Experimental Settings." American Economic Review 80:902-11.

Corrigan, J., and M. Rousu. 2006. "The Effect of Initial Endowments in Experimental Auctions." American Journal of Agricultural Economics 88:448-57.

Cox, J., and D. Grether. 1996. "The Preference Reversal Phenomenon: Response Mode, Markets and Incentives." Economic Theory 7:381-405.

Cubitt, R., and R. Sugden. 2001. "On Money Pumps." Games and Economic Behavior 37:121-60.

Ehrenberg, R., and M. Bognanno. 1990. "Do Tournaments Have Incentive Effects?" Journal of Political Economy 98:1307-24.

Gode, D., and S. Sunder. 1993. "Allocative Efficiency of Markets with Zero-Intelligence Traders: Market as a Partial Substitute for Individual Rationality." Journal of Political Economy 70:119-37.

Gunnarsson, S., J. Shogren, and T. Cherry. 20(13. "Are Preferences for Skewness Fixed or Fungible?" Economics Letters 80:113-21.

Hayek, F. 1945. "The Use of Knowledge in Society." American Economic Review 35:519-30.

Heiner, R. 1983. "'The Origin of Predictable Behavior." American Economic Review 76:560-95.

Henrich, J., R. Boyd, S. Bowles, C. Camerer, E. Fehr, H. Gintis, R. McElreath, M. Alvard, A. Barr, J. Ensminger, N. Smith Henrich, K. Hill, E Gil-White, M. Gurven, F. Marlowe, J. Patton, and D. Tracer. 2005. "'Economic Man' in Cross-Cultural Perspective: Behavioral Experiments in 15 Small-Scale Societies." Behavioral and Brain Sciences 28:795-855.

Hoffman, E., K. McCabe, and V. Smith. 1996. "Social Distance and Other-Regarding Behavior in Dictator Games." American Economic Review 86:653-660.

Horan, R., E. Bulte, and J. Shogren. 2005. "How Trade Saved Humanity from Biological Exclusion: An Economic Theory of Neanderthal Extinction." Journal of Economic Behavior & Organization 58:1-29.

Hsee, C., and R. Hastie. 2006. "Decision and Experience: Why Don't We Choose What Makes Us Happy." Trends in Cognitive Sciences 10:31--37.

Kahneman, D., and A. Tversky, eds. 2000. Choices, Values and Frames. Cambridge: Cambridge University Press.

Knetsch, J. 1989. "The Endowment Effect and Evidence of Nonreversible Indifference Curves." American Economic Review 79:1277-84.

Kreps, D., F. Milgrom, J. Roberts, and R. Wilson. 1982. "Rational Cooperation in the Finitely Repeated Prisoners' Dilemma." Journal of Economic Theory 27:245-52.

Lichtenstein, S., and F. Slovic. 1971. "Reversals of Preferences between Bids and Choices in Gambling Decisions." Journal of Experimental Psychology 101:16-20.

Loomes, G., S. Starmer, and R. Sugden. 2003. "Do Anomalies Disappear in Repeated Markets." Economic Journal 113:C153-66.

Lynne, G. 20(/6. "Toward a Dual Motive Metaeconomic Theory." Journal of Socio-Economics 35:634-51.

McFadden, D. 1999. "Rationality for Economists?" Journal of" Risk and Uncertainty 19:73-105.

McKelvey, R., and T. Palfrey. 1992. "An Experimental Study of the Centipede Game." Econometrica 60:803-36.

Menkhaus, D., G. Borden, G. Whipple, E. Hoffman, and R. Field. 1992. "An Empirical Application of Laboratory Experimental Auctions in Marketing Research." Journal of Agricultural and Resource Economics 17:44-55.

Nau, R., and K. McCardle. 1991. "Arbitrage, Rationality, and Equilibrium." Theory and Decision 31: 199-240.

Ofek, H. 2001. Second Nature: Economic Origins of Human Evolution. Cambridge: Cambridge University Press.

Parkhurst, G., and J. Shogren. 2005. "Does Complexity Reduce Coordination?" Applied Economic Letters 7:447-52.

Pinker, S. 2003. The Blank Slate: The Modern Denial of Human Nature. New York: Penguin.

Plott, C. 1994. "Market Architectures, Institutional Landscapes and Testbed Experiments." Economic Theory 4:3-10.

Ridley, M. 2003. Nature via Nurture: Genes, Experience, and What Makes Us' Human. New York: Harper-Collins.

Schelling, T. 1960. The Strategy of Conflict. London: Oxford University Press.

Shogren, J., and T. Crocker. 1991. "Risk, Self-Protection, and Ex Ante Economic Value." Journal of En vironmental Economics and Management 21:1-15.

Shogren, J., G. Parkhurst, and C. McIntosh. 2006. "Second Price Auction Tournament." Economics Letters 92:99-107.

Slovic, F. 2000. "The Construction of Preferences." In D. Kahneman and A. Tversky, eds. Choices, Values, and Frames'. Cambridge: Cambridge University Press, pp. 489-502.

Smith, V.L. 2003. "Constructivist and Ecological Rationality in Economics." American Economic Review 93:465-508.

Thaler, R., and C. Sunstein. 2003. "Libertarian Paternalism Is Not an Oxymoron." University of Chicago Law Review 70:1159-202.

Tversky, A., and D. Kahneman. 2000. "Rational Choice and the Framing of Decisions." In D. Kahneman and A. Tversky, eds. Choices, Values, and Frames. Cambridge: Cambridge University Press, pp. 209-23.

Tversky, A., and I. Simonson. 1993. "Context-Dependent Preferences." Management Science 39:1179-89.

Vickrey, W. 1961. "Counterspeculation, Auctions, and Competitive Sealed Tenders." Journal of Finance 16:8-37.

Waugh Lecture.

Jason E Shogren is the Stroock Distinguished Professor of Natural Resource Conservation and Management, Department of Economics and Finance, University of Wyoming, Laramie, WY.

Thanks to the ERS/USDA for supporting my research. Thanks to the seminar participants who helped me sharpen this story for many years now, the editor, and to my colleagues for sharing their great ideas, Tom Crocker, Todd Cherry, Sean Fox, Stephan Kroll, to name a few. Thanks to my friend Dermot Hayes for wondering out loud what experimental economics might have to contribute to agricultural economics. Finally, thanks to the Norwegian University of Life Sciences for the hospitality needed to finish this paper.

This article was presented as the Waugh Memoral Lecture at the AAEA Annual Meeting of the Amencal Agriculture Economics Association in Long Beach, CA, July 2006. Invited addresses are not subject to the journal's standard refereeing process.

Table 1. Play of Payoff Dominant
Strategy-Individual

         Normal Form     Grid Game

          Payoff          Payoff
Rounds   Dominant   N    Dominant   N

1-5       82.5%     80     35.0%    80
6-10      90.0%     80     88.8%    80
11-15     90.0%     80    100.0%    80
16-20     93.8%     80    100.0%    80

Figure 2. Normal form payoff matrix

                      Player A

Player B      1        2         3        4

1           60 60    60 105    60 95    60  85
2          105 60   105 105   105 95   105  85
3           95 60    95 105    95 95    95  85
4           85 60    85 105    85 95   135 135
                                       Nash Eq
5          111 60   111 105   111 95   111  85

6          109 60   109 105   109 95   109  85
7          101 60   101 105   101 95   121 105

8           99 60    99 105   99  95   129 115

                      Player A

Player B         5         6         7         8

1           60 111    60 109    60 101    60  99
2          105 111   105 109   105 101   105  99
3           95 111    95 109    95 l0l    95  99
4           85 111    85 109   105 121   115 129

5          111 111   111 109   111 101   111  99
           Nash Eq
6          109 111   109 109   109 101   109  99
7          101 111   101 109   121 121   121 119
                               Nash Eq
8           99 111    99 109   119 121   129 129
                                         Nash Eq

Figure 4. Dictator within context: how would you split $100?

                          Scenario A:         Scenario B:
                          You did not earn    You Earned the
The other person is a     the $100 (e.g., a   $100 (e.g., waiter,
complete stranger         gift, found it)     bartender)

The other person has      $__ You             $__ You
not had the opportunity
to earn any money
                          $__ Other person    $__ Other person

The other person has      $__ You             $__ You
not earned any money
                          $__ Other person    $__ Other person

The other person has      $__ You             $__ You
decided not to earn
extra money               $__ Other person    $__ Other person