In our previous primer on neoclassical (traditional) and behavioral economics,1 we predominately focused on neoclassical economic principles as they pertain to transplantation. In part 1, we briefly expanded our discussion to highlight some behavioral economic principles. This primer will primarily focus on a few additional behavioral economic principles that we find relevant to the individual decisions made in transplantation with a few references to how they originated from anomalies discovered using neoclassical economic theory. Our reasoning for this ordering follows the development of economic theory.
The fundamental purpose of economic theory is to develop models that can be used to predict individual decision making. Therefore, the success of our models rests on their ability to accurately predict individual decisions. Economics as a discipline incorporates human psychology and makes assumptions about what constitutes a rational decision making agent.2 The field of behavioral economics arose from an interest in revisiting a number of key assumptions that define a “rational” human often referred to as homo economicus,3 and determining whether or not those assumptions are valid. It also focuses on whether understanding certain underlying fallacies in human rationality can expand our existing economic theories to develop models that better predict individual decisions. In this way, behavioral economics is not an alternative to neoclassical economics but more of an evolutionary step in our scientific understanding of human decision making. This is best summarized in Camerer and Loewenstein's statement,
“All economics rests on some sort of implicit psychology. The only question is whether the implicit psychology in economics is good psychology or bad psychology. We think it is simply unwise, and inefficient, to do economics without paying some attention to good psychology…..Our hope is that behavioral models will gradually replace simplified models based on stricter rationality, as the behavioral models prove to be tractable and useful in explaining anomalies and making surprising predictions. Then strict rationality assumptions now considered indispensible in economics, will be seen as useful special cases.” (Camerer and Loewenstein,2 p 42).
The selected topics from behavioral economics that we have elected to discuss in this primer are pertinent to the decisions made daily in transplantation and include fairness and “other regarding preferences,” framing effects, the law of small numbers, and hyperbolic and quasi-hyperbolic discounting. However, it is worth noting that there are a number of other behavioral economic principles that relate to these topics that may be of interest which are discussed in the book Advances in Behavioral Economics,4 as well in the recent literature by the researchers who contributed to the book. It is our hope that through this primer and our superficial discussion of the applicability of these concepts to organ transplantation that the reader will better understand the individual decisions made by recipients, donors, surgeons, and others in the transplant community. A better understanding of these microdecisions will allow us to more precisely evaluate how our public policies influence the organ transplant process.
OTHER REGARDING PREFERENCES
The neoclassical utility function illustrated in our previous primer1 assumed that an individual solely makes decisions based on their own self-interest. However, if one is asked whether or not self-interest is their only motive for a decision, most would indicate that there are other factors that influence their decision. For instance, we may be motivated by fairness, equity, reciprocity, or altruism. These motivations do not necessarily invalidate the neoclassical model; instead they suggest that this model needs to be expanded to incorporate both self-interest as well as preferences that expand the traditional view of an economic human being. This is achieved through the consideration and development of other regarding preferences, which introduces a higher degree of psychological realism to the traditional self-interest model.5 A well-cited example of the need to expand our neoclassical utility function comes from the ultimatum game.6 In the ultimatum game, there are 2 people splitting an arbitrary amount of money, which we will normalize to be US $1 for illustration. The first mover decides how much of the US $1 to pass along to the second mover, call it X. The second mover observes X and then can elect to accept X or not accept X. If the second mover accepts X then the first mover receives US $1 minus X, and the second mover receives X. If the second mover does not accept X, they both will receive a payoff of zero. Neoclassical economic theory would predict that the first mover will never pass more than 1 cent, and the second mover will accept anything greater than 1 cent.7 However, this is not what happens when this experiment is conducted. Instead, it is often the case that a very large percentage of the offers made by the first mover are between 40% and 50% of the principal value, and very few offers are below 20% of the principal value.7 As discussions regarding organ allocation take place, it would be useful to keep in mind that differing points of view are not purely self-interest, but rather complex behaviors combining elements of both self-interest and other regarding preferences, such as fairness and altruism.
A number of models have been developed to expand the neoclassical utility function to incorporate other regarding preferences.7-9 A central feature of these models is that one's preferences include not only their interests but also the interests of others who are impacted by their decision. Furthermore, a number of these models are consistent with the hypothesis of behavioral reference points and loss aversion,10 as the reference point is often the returns that others derive and loss aversion arises when one's returns are lower than another party's return.7
As mentioned previously, an area where concerns of “fairness” have played a sizeable role in transplantation is the organ allocation process as regulators have altered the ordering of offers (for example, Share 3511) and are considering redefining the spatial boundaries of the organ allocation districts to address issues of fairness.12Figure 1 graphically illustrates preferences that exhibit a tradeoff between one's own return and the return of another. This figure is an adaptation of a figure originally appearing in Bolton and Ockenfels9 and based on their concept of a “motivation function.”9 These preferences are contextualized in terms of regional organ allocation. For simplification, we have assumed that there are only 2 regions being considered, regions 1 and 2, for the allocation of organs, and our preferences illustrate those of region 1. In this setting, the motivation function, analogous to a value function, increases as the share of organs allocated to region 1 increases. Furthermore, the marginal increases in the motivation function are large when the percentage of organs allocated to region 1 is low and the marginal gains decrease when the allocation increases. Lastly, as the percentage of organs allocated to region 1 begins to exceed approximately 65%, implying 35% allocated to region 2, the motivation function begins to fall; the marginal gains are negative. This is because the decision agent would prefer to have the additional share of organs allocated to region 2 versus region 1 because adding more to region 1 reduces the benefits they derive—allocating too large of a percentage to region 1 would be “unfair.” The application of these behavioral theories can be used to derive the measures of fairness exhibited by the regulatory bodies that oversee organ allocation among the different transplant regions within the United States.
The discussion of Prospect Theory10 in our first primer illustrated that an individual's preferences may depend on how the decision environment is framed (ie, whether the change being valued is perceived as a gain or loss). The concept of framing effects suggests that one's preferences may reverse depending on how the information is framed. A classic example of framing effects is Tversky and Kahneman's disease-risk example.13 In this example, subjects are informed that the United States is preparing for a disease epidemic that is expected to kill 600 people, and there are 2 alternative programs that can be considered to address the health concern. These 2 alternatives are framed in 2 different ways. In the first framing subjects are asked to select between 1 program that will save 200 people with certainty (option A) and a second program that if adopted will result in a one third probability of all 600 being saved and a two third probability that no one will be saved (option B). Under this framing, a majority of the subjects (72%) selected the first option (option A).13 This option is consistent with risk aversion because subjects would rather save 200 lives with certainty than risk an uncertain option that generates 200 expected lives saved.13 Under the alternate framing, the first program stated that it would result in 400 people dying from the disease (option C), and the second program contained a one third probability that no one would die and a two third probability that 600 people would die (option D). Under this alternate framing, a majority of the subjects (78%) selected the second option (option D).13 This result indicates that subjects are risk loving because they would rather risk losing all 600 people to the disease versus losing 400 with certainty even when the expected number of lives lost under the second alternative is 400.13 Therefore, under 1 framing, subjects exhibit risk aversion, and under another, they exhibit risk loving behavior. This arises even though under both of the framings the expectations are the same. The only difference is how the decision is framed. This behavior is consistent with the Prospect Theory10 model discussed in part 1 of this primer.1
Within transplantation, there are a number situations in which the framing of a decision may influence people's preferences. One example is the discussion of financial incentives for organ donation versus reducing the financial disincentives of being an organ donor. Providing financial compensation for organ donation generates what has been described as a “morally repugnant” market because it objectifies a good via monetary value—in this case, an organ—there exists a potential for coercion and exploitation of the poor, and we create a slippery slope when monetizing a particular good.14 This sense of moral repugnance is consistent with the considerable ethical considerations that have arisen as reasons for opposing incentives for organ donation.15-19 Regardless, it is obvious that the current mechanisms used to reduce the financial disincentives are insufficient to meet the growing demand for organs. Therefore, we are left to hypothesize what would be a rational middle ground between these 2 states. On the one hand, we can use a market to meet the demands of transplantation, which would come at a potential cost of exploiting the impoverished. Alternatively, we can continue to use minimal financial incentives that are insufficient to incentivize donation and continue to have an increasingly large number of patients who die while waiting for an organ transplant. The costs to society are large under both options. Further, as we move more and more to patient satisfaction and other qualitative metrics, framing of the tools used may have a profound impact on these metrics.
Glossary of Terms from Primer 1 and Primer 2
Discounting: Evaluating the value of future returns using its current value so that all decisions made regarding current and future returns are valued using the same metric.
Disincentives: The cost of taking an action that may prevent an individual from taking that action.
Framing: The manner in which a choice is presented to an individual, which may influence their choice. This may generate preference reversals that are inconsistent with other choices that an individual has made.
Indifference Curve: Is a graphical representation of a curve that indicates all bundles of consumption or states of nature that provide the same level of utility for an individual.
Marginal Rate of Substitution: A measure of one’s willingness to trade one good or state of nature for another while maintaining the same level of utility.
Other Regarding Preferences: Preferences that span beyond traditional self-interest. This includes one’s desire for fairness, equity, reciprocity and altruism. These preferences can be incorporated into one’s utility function.
Utility: A measure of individual satisfaction assigned to a bundle of goods consumed or a state of nature. It is an ordinal measure used to compare an individual’s satisfaction across different levels of consumption or states of nature.
Behavioral economics would inform us that the manner in which this tradeoff is presented may influence one's preferences. For instance, by using the phrase “reduced financial disincentives” versus “increased financial incentives” to motivate this discussion, we are directly altering the lens through which the decision is viewed. Will society be more willing to support the use of incentives for organ donation if it is framed as a “reduced disincentive” versus an “increased incentive”? This is a behavioral economics question, but the current approach in transplantation is to frame it as a “reduced disincentive,” which we hypothesize is to ensure broader support. It directly implies that those who are current donors are worse off than if they did not donate, whereas the “increased incentives” are not. In essence, this framing effect generates a behavioral reference point. A “reduced disincentive” implies a loss from a reference point that is an individual's welfare before deciding to donate, whereas an “increased incentive” implies a gain from a reference point that is an individual's welfare after they have elected to donate. This is an important distinction, and it may explain why the recent literature in transplantation has begun to focus more on the concept of reducing the financial disincentives of being an organ donor.20,21
THE “LAW OF SMALL NUMBERS”
After our discussion of the neoclassical utility function in our first primer,1 we outlined how this model can be applied to decisions under uncertainty. As part of this discussion, we assumed that rational decision agents accurately process information on probabilities. The “law of small numbers,” first coined by Tversky and Kahneman,22 challenges this assumption. The law of small numbers is a behavioral phenomenon where individuals view a small random sample of the population as representative of the population at large. An example of this is what is referred to as the “gambler's fallacy.”22 Suppose one observes a sequence of 2 coin flips that are both heads, one who suffers from the gambler's fallacy believes that the probability of the third coin flip coming up tails is greater than 0.50, despite the fact that the coin flips are independent draws. This is the law of small numbers, as individuals become prone to believe that there is a self-correcting process that will restore the observed small sample to what one would observe in a large sample.22 This is best summarized in the following statement made by Tversky and Kahneman, “some familiar processes in nature obey such laws: a deviation from a stable equilibrium produces forces that restore the equilibrium. The laws of chance, in contrast, do not work that way: deviations are not canceled as sampling proceeds, they are merely diluted.”22, p.106
Rabin23 expanded the law of small numbers concept developed by Tversky and Kahneman.22 Rabin23 illustrates that individuals possessing beliefs consistent with the law of small numbers are not only more likely to overestimate the expected balancing of small sequences to represent the larger population, but they are also more likely to suffer from “overinference” when a series of small strings are observed by a decision agent, and they are uncertain about the underlying distribution. Rabin23 uses this model of overinference to explain why in the stock market individuals may be more prone to overestimate the probability that a mutual fund manager who is outperforming the market in recent years will continue to do so despite the fact that the underlying performance distribution does not support this prediction.
The “law of small numbers” is highly relevant in organ transplantation because each individual transplant center conducts a small sample of transplants drawn from the set of national transplants that defines the underlying distribution. The national statistics are used by the Centers for Medicare and Medicaid Services to monitor the performance of centers using 1-year graft and patient survival metrics. As a result, centers must closely monitor their current performance and make decisions regarding whether or not conducting a particular transplant will put them at risk for Centers for Medicare and Medicaid Services review. If the center makes decisions consistent with the “law of small numbers,” this would imply 2 important generalizations regarding their perception of the previous transplants conducted. First, if their most immediate transplants resulted in a higher level of quality relative to the national standards, they would anticipate that the next transplant may cancel out the gains obtained via a reduction in observed quality. Second, if they continued to obtain successful transplants, thus generating some uncertainty regarding the underlying transplant distribution, they will begin to start to overestimate the probability that the string of good outcomes will continue. This is referred to as “overinference” by Rabin23 and could have damaging effects on a center's performance because they may take on extra risk that cannot be rationalized given the national distribution of transplants. This may generate a reduction in their 1-year graft and patient survival rates and compromise their regulatory status.
At this juncture, we have only tangentially spoken of how individuals make decisions over time by highlighting it is as an important component of a surgeon's utility function when deciding to accept or reject an organ.1 Given that many of the decisions recipients, donors, and surgeons make possess a strong temporal component, it is imperative that we incorporate this topic within our primer of neoclassical and behavioral economics.
Neoclassical economics uses discounted utility to model intertemporal preferences and choice.24 The discounted utility model decomposes the decisions made each period into separable utility functions and then links them using a discount function that evaluates all future utility in terms of its current value.24 For instance, a US $100 today is worth more than US $100 a year from now because the current money could be invested and earn interest. Therefore, the US $100 earned next year would be discounted to an equivalent value today such that if it were invested today it would generate US $100 in 1 year. This same process is conducted for the utility that one derives from taking a specific action in the future. The discounted utility model sums all of these discounted utilities and evaluates them in current value terms. Therefore, this model conveniently expands our utility model to one of intertemporal choice via the addition of 1 new parameter, the discount factor.24,25
Although the discounted utility function is mathematically tractable, it is definitely one of the biggest “straw men” in the neoclassical economics toolbox as many of its simplifying assumptions have been shown to be invalid. Behavioral economists have illustrated its inability to predict a number of behavioral anomolies26 and proposed alternative models that more accurately reflect observed behavior.27,28 Two alternative models that have come from the behavioral economics literature are the hyperbolic discounting27 and quasi-hyperbolic discounting28 models. The central feature of these discounting functions is that they both assign a higher value to utility that is derived in recent periods followed by a much lower and decreasing value in future periods. The difference between the 2 is that the hyperbolic is a smoother discounting function, whereas quasi-hyperbolic possesses a kink at the point where recent values become future values. Both of these models imply that individuals are much more likely to focus on immediate gains versus delaying them in the future, and this will result in dynamically inconsistent decisions that violate the assumptions of discounted utility. Dynamic inconsistency implies that if one were to make a decision today regarding a reward they were to receive in X periods in the future, they would not make the same decision once they arrive at X periods in the future. A benefit of the discounted utility model is that it does not generate dynamically inconsistent behavioral predictions. Figure 2 illustrates the differences between the exponential discounting assumed under discounted utility24 as well as hyperbolic27 and quasi-hyperbolic28 discounting.
Discounting is extremely important in transplantation because many of the decisions made by recipients, donors, and surgeons are intertemporal decisions. For instance, at the time a patient is being listed for transplant, they may be asked if they are willing to accept an organ from an extended criteria donor (ECD). Controlling for the patient's health status, this decision is based on the patient's expectations of the length of time that the graft will survive and provide an improved quality of life. If the patient exhibits hyperbolic or quasi-hyperbolic discounting, they will most likely focus on the immediate gains obtained from the ECD organ. This is not to suggest that these preferences are somehow “incorrect,” it merely suggests that if we wish to develop models that predict when a patient is willing to accept an ECD organ that we correctly use the appropriate discounting model.
Another area in transplantation where these preferences may be important is with postoperative care. After a transplant, the success of an organ is in many ways driven by the health decisions that recipients make after leaving the hospital. Making good health decisions that will improve the success of an organ will often times require recipients to make decisions today that will provide health benefits in the future. If a recipient exhibits hyperbolic or quasi-hyperbolic preferences, they may be less willing to take actions that reduce their instantaneous utility to increase their health in the future; this may compromise their future health as well as their graft survival rate. Furthermore, if recipients are required to commit to a path of health decisions (ie, drug therapies, follow-ups, and so on), the dynamic inconsistency of their decisions would imply that they are less willing to actually take these actions when the day comes for them to take the precommitted action. Knowledge of this may help inform the mechanisms used to follow-up with recipients after a transplant to better improve graft and patient survival rates. For instance, a recipient who exhibits either hyperbolic or quasi-hyperbolic discounting is more willing to precommit to a path that constrains future choices.28 If these precommitment mechanisms are binding, it may in fact be a mechanism to improve a recipient's health decisions after transplantation.
Our discussion of the neoclassical and behavioral economic theories outlined in this primer, as well as our previous primer,1 provide a number of decision-making frameworks for us to use to critically evaluate individual decisions made in organ transplantation. It is our hope that this discussion has provided a number of alternative decision-making frameworks that researchers can use to analyze the decisions made by recipients, donors, surgeons, and others in the transplant community. The application of these models will require detailed knowledge of the transplant environment as well as these economic theories. Therefore, there are a plethora of potential future collaborations between social scientists and clinical researchers in transplantation that can advance public policy. We hope that our discussion has stimulated enough interest among these groups to develop further scientific advancements in the study of individual decision making in organ transplantation.
1. Schnier KE, Turgeon N, Kaplan B. A primer of neoclassical (traditional) and behavioral economic principles for organ transplantation: part 1. Transplantation
. 2015; 99.
2. Camerer CF, Loewenstein G. Behavioral Economics: Past, Present, Future. In: Camerer CF, Lowenstein G, Rabin M, editors. Advances in Behavioral Economics
. Russell Sage Foundation, New York, New York: Princeton University Press, Princeton and Oxford; 2004: 2–51.
3. Thaler RH. From Homo Economicus to Homo Sapiens. J Econ Perspect
. 2000; 14: 133–141.
4. Camerer CF, Loewenstein G, Rabin M. Advances in Behavioral Economics
. Russell Sage Foundation, New York, New York: Princeton University Press, Princeton and Oxford; 2004.
5. Rabin M. An Approach to Incorporating Psychology into Economics. Am Econ Rev
. 2013; 103: 617–622.
6. Thaler R. Anomalies: The Ultimatum Game. J Econ Perspect
. 1988; 2: 195–207.
7. Fehr E, Schmidt KM. A theory of fairness, competition, and cooperation. Q J Econ
. 1999; 114: 817–868.
8. Rabin M. Incorporating fairness into game theory and economics. Am Econ Rev
. 1993; 83: 1281–1302.
9. Bolton GE, Ockenfels A. ERC: a theory of equity, reciprocity, and competition. Am Econ Rev
. 2000; 90: 166–193.
10. Kahneman D, Tversky A. Prospect theory: an analysis of decisions under risk. Econometrica
. 1979; 47: 263–291.
11. Gentry SE, Chow EK, Massie A, et al. Liver sharing and organ procurement organization performance. Liver Transpl
. 2015; 21: 293–299.
12. Gentry SE, Massie AB, Cheek SW, et al. Addressing geographic disparities in liver transplantation through redistricting. Am J Transplant
. 2013; 13: 2052–2058.
13. Tversky A, Kahneman D. The framing of decisions and the psychology of choice. Science
. 1981; 211: 453–458.
14. Roth AE. Repugnance as a constraint on markets. J Econ Perspect
. 2007; 21: 37–58.
15. Pellegrino ED. Families' self-interest and the cadaver's organs. What price consent? JAMA
. 1991; 265: 1305–1306.
16. Delmonico FL, Arnold R, Scheper-Hughes N, et al. Ethical incentives—not payment—for organ donation. N Engl J Med
. 2002; 346: 2002–2005.
17. Israni AK, Halpern SD, Zink S, et al. Incentive models to increase living kidney donation: encouraging without coercing. Am J Transplant
. 2005; 5: 15–20.
18. Steinbrook R. Public solicitation of organ donors. N Engl J Med
. 2005; 353: 441–444.
19. Rothman DJ. Ethical and social consequences of selling a kidney. JAMA
. 2002; 288: 1640–1641.
20. Delmonico FL, Martin D, Domínguez-Gil B, et al. Living and deceased organ donation should be financially neutral acts. Am J Transplant
. 2015; 15: 1187–1191.
21. Fisher JS, Butt Z, Friedewald J, et al. Between Scylla and Charybdis: charting an ethical course for research into financial incentives for living kidney donation. Am J Transplant
. 2015; 15: 1180–1186.
22. Tversky A, Kahneman D. Belief in the law of small numbers. Psychol Bull
. 1971; 76: 105–110.
23. Rabin M. Inference by believers in the law of small numbers. Q J Econ
. 2002; 117: 775–816.
24. Samuelson P. A note on measurement of utility. Rev Econ Stud
. 1937; 4: 155–161.
25. Frederick S, Loewenstein G, O’Donoghue T. Time discounting: a critical review. J Econ Lit
. 2002; 40: 351–401.
26. Thaler R. Some empirical evidence of dynamic inconsistency. Econ Lett
. 1981; 8: 201–207.
27. Loewenstein G, Prelec D. Anomalies in intertemporal choice: evidence and an interpretation. Q J Econ
. 1992; 107: 573–597.
28. Laibson D. Golden eggs and hyperbolic discounting. Q J Econ
. 1997; 112: 443–477.