The Impact of Default Rules on Economic Behavior, With Primary Attention to Organ Donations
Sebastien R. Gay1 - University of Chicago
January 2006
Abstract
This paper discusses the impact of default rules on economic behavior. We primarily focus on the effect of legislation on cadaveric organ donations. We show how the involvement of the family in the decision making process switches the common belief that the opt-out system (donors by default) results in more organ donations than the opt-in system (non-donors by default). Informed consent systems could actually result in more donations than presumed consent. We use Sweden, a country that switched their system twice over the years, and a separate panel data of 15 countries spanning over the period 1990-2002, to test our model.
JEL Classification: D19, I18
1
Sebastien R. Gay is a Graduate Student at the Department of Economics of the University of Chicago, 1126 East 59th Street, Chicago, IL 60637 (
[email protected]) I would like to thank Alberto Abadie and Philippe Fevrier for their support and help. I also thank Gary Becker, Robert Topel, Steve Levitt, Hugo Sonnenschein and Pierre-Andre Chiappori for their guidance and helpful comments. All remaining errors are mine.
I. Introduction
The existence of defaults impacts the economic behavior of households. Switching to a new telecom provider, changing banks, enrolling in savings plans, donating organs are examples of decisions that often result from a “no-decision”, i.e. staying in a default situation. Surprisingly these “no-decisions” do not seem to have a high cost for the individuals but in practice, if no action is taken, the outcome can turn out to be costly, especially in the long run: pay higher phone bills, higher costs or higher interest rates on credit cards at current bank, not enough savings later in life, death of people on the waiting lists. In theory economic agents are defined as rational agents that solve a maximization program to find the optimal quantities of goods they want to consume. Nonetheless, even if not satisfied by the end result, individuals often stay in default situations: for example, a study in Consumer Reports Magazine (2005) reports that customers' overall satisfaction with their cell phones is much lower than for any other industry but only 25% of the customers actually change carriers at the end of their contracts, whereas more than 70% of them declare wanting to switch carriers. The intriguing fact is that the costs of switching are usually low, as closing a bank account and changing carriers for cell phones at the end of a contract are an easy and almost costless procedure (especially with the generalized portability of numbers). Hence the explanation seems to lie beyond the usual theory. Such a behavior is displayed when individuals are faced with a problem whose necessity does not lie in the present and a switch now does not seem to have important benefits in the future (or benefits are so far remote that they are disregarded by the individuals): agents hence prefer to stay in the default situation they are placed in.
The essence of such defaults relies on status quo or procrastination of individuals: when placed in a default situation or situation of reference, individuals rarely want to switch and instead stay in the default rule. This so called status quo has been proved to be central even when strong reasons for change exist. Loss aversion, anticipated regret and ignorance (Kahneman and Tversky (1979); Kahneman, Knetch and Thaler (1991); Laibson (1998); Samuelson and Zeckhauser (1988)) are among the explanations that cause inaction. Also, procrastination is to keep delaying something that must be done, often because it is unpleasant. Procrastination has been studied in economics in order to better understand inadequate savings or organizational failures (Akerlof 1991). Recent papers have emphasized the importance of procrastination in a default system for savings: Choi et
2
al. (2003) and Thaler and Benartzi (2004) show that default options have a remarkable impact on household choices in term of savings, for opting out of a default is costly and people’s tendency to procrastinate. Savings levels indeed do not match the optimal level for retirement for most individuals: the Employee Benefits Research Institute2 (EBRI) reports that in 2005 only less than 40% of the US workers have calculated how much they need to retire, 30% have not saved anything for retirement, only 20% are very confident about having enough money to live a good life in retirement. Choi et al (2001) use survey data to understand savings patterns and distinguish what people should do and what they actually do. The following Table 1 reproduces the findings of Choi et al (2001) in terms of planned change and actual change in their sample. One of the most stunning figures is the willingness of agents to decide to enroll in a 401(k) because they know that it will contribute to their retirement savings, but only 14% of them have finally enrolled in a 401(k).
Action
Planned Change
Actual Change
Enroll in 401(k)
100%
14%
Increase Contribution Rate
28%
8%
Change fund selection
47%
15%
Change fund allocation
36%
10%
Table 1: Planned versus Actual Change: Do Agents Follow their Intents?
Madrian and Shea (2001) also study the effect of the default system on savings. They found that 86% of the employees hired after the implementation of an automatic participation rule stay in the default savings plan designed by the company (compared to 37% enrollment before the implementation of the new plan). Defaults have sometimes the drawbacks of their own successes: Madrian and Shea (2001) also report that the average savings rate decreased from 7.2% with no automatic enrollment to 4.4% under automatic enrollment, because the automatic enrollment had a default rate of 3% (hence even workers who might have chosen a higher plan “decided” to take the automatic plan with 3% rate). This shows the potential problems of default rules: more workers in numbers save money for retirement, but to attract their employee in the default plan and decrease the probability of opting out of the plan, companies decide to have a savings rate inferior to the interest rate without automatic enrollment (and usually under the optimal rate). Hence setting a default is a quality2
The 2005 EBRI Report on savings is available at http://www.ebri.org/pdf/PR_692_5Apr05.pdf
3
quantity adjusted issue: does a company want more savings plans in quantity at a lower rate or only quality savings plan with a higher rate?
This paper considers similar consequences of defaults on other economic behavior, focusing primarily on organ donations. There are two types of default legislation on cadaveric organ donations: presumed consent (PC) and informed consent (IC). In informed consent countries, also called opt-in countries, individuals are only donors when deceased if they registered to do so while alive. This system is mostly used in countries like the United States, Canada, New Zealand, Great Britain and Australia. Conversely, in presumed consent countries, also called opt-out countries, anybody is potentially a donor by default. People have to register if they do not want to donate their organs. European countries (except Germany, the Netherlands, UK and Ireland) usually apply this legislation. In the last decade the debate on organ donations has primarily focused on legislation3, with the well accepted argument that a switch from informed to presumed consent should increase organ donations (Spital (1996) and Mustarah (1998), Johnson and Goldstein (2003)). The intuition behind this result is that a presumed consent system is likely to benefit from the organs of potential donors that did not declare any preference for donation while living and hence would stay in the default situation.
Figure 1 shows the evolution of organ donations according to each type of legislation. Abadie and Gay (2004) show using panel data on 22 countries over the period 1993-2202 that presumed consent legislation brings 25% to 30% more organ donations than informed consent, controlling for other determinants4.
We will prove that surprisingly, and contrary to the usual belief, the informed consent default could lead to more donations if the will of the family of the deceased is taken into account. We then test our model on Sweden, which changed its legislation twice between 1981 and 2005. We finally test the conclusions of our model on a panel data of 15 countries with official organ donation registers. Our empirical estimation is consistent with the theory developed in the model.
3
Another debate concerns the use of monetary incentives to increase the supply of organs (Cohen, 1989; Becker, 1997; Becker and Elias, 2003). 4 This paper gives a clear insight on the organ donation empirical issues, although intrinsic structural differences between countries could potentially bias the result.
4
Figure 1: Evolution of Organ Donations According to the Legislation
24
22
Donations in PMP
20
18 Informed Consent Presumed Consent 16
14
12
10 1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
Years
Note: The informed consent line was derived from the donation rates of the following seven countries: Australia, Denmark, England, Germany, the Netherlands, USA and Switzerland. The presumed consent line was derived from the donation rates of the following seven countries: Austria, Belgium, France, Italy, Poland, Portugal and Spain.
This paper is organized as follows. Part II discusses the importance of defaults on other sectors of the economy, like telecommunications. Part III focuses on organ donations and shows the theoretical impact of the family on organ donation depending on the default legislation. Part IV considers the Swedish case. Part V tests the importance of presumed consent on organ donations for countries with registers. Part VI concludes.
II. Telecommunications and Defaults: Switch or Status Quo?
In 1998 the French government decided to break the monopoly of France Telecom as a provider of telecommunication services to comply with the free market policy promoted by the Treaty of the European Union5. It first started with long distance calls in January 1998, and then by 2002, local calls were also deregulated.
5
Until 31st December 1997: France Telecom had the absolute monopoly on telephone communications, like EDF for electricity or SEITA for cigarettes and matches.
5
Technically customers could switch very easily to a new provider: if they wanted to use another provider of long distance phone calls, they are only required to start dialing the phone number they want to reach with a different number (between 2 and 9 depending on the operator) instead of the “0” of France Telecom in front of the other seven numbers making up the correspondent’s number. Note that this procedure was free of charge.
We consider figures reported by the French Authority of Regulation of Telecommunications (ART) taken from a sample of 2076 households in order to measure the persistence of defaults between 1998 and 2002 (the default here is to stick to France Telecom) 6 . The prices of the different operators were lower than those of France Telecom (Table 2a shows indices of the evolution of the prices of telecommunications) and the calling quality was considered the same by the ART (1998) even at the beginning of deregulation. In January 1998, 34.1 million people were clients of France Telecom, and at the end of December 1998, 860,000 chose to be clients of a different operator. In December 1999, only 2,964,098 were clients of another phone company, i.e. only 5.88% of the former France Telecom customers switched after 2 years. In January 2003, out of the 86% of French households with a ground line7, 64% were still exclusive customers of France Telecom, with 19% being clients of another company.
We see that this trend is actually in Europe, as most of the clients stay with the former national monopoly, but at different degrees: in 2004, 75% of clients in the telecommunications sector in France, Ireland, and the Netherlands stayed with their national monopoly. On the contrary, in the UK, only 50% are clients of British Telecom, and 60% of Deutsche Telecom in Germany8.
The striking result is that the switch in phone providers has been massively operated by companies or local governments: ART (2001) shows that the switch of operators has been operated by local governments (75% of cities or counties were clients of another company in
Since 1st January 1998: You can opt-in for any phone operator, only for long distance and international communications (i.e. further than the department). 9 Telecom and Cegetelto provide this service. 1st November 2000: Market for communications to mobile phones was also opened to competition. Since January 2002: Entire market was opened to competition, local calls included (the only mandatory fee the 13 euro-subscription for the line!). 6 In 1998, 98% of French households had a ground line and the use of cell phones only started to become rampant in 2000, providing us with interesting comparisons avoiding the substitution effect between cell phone and ground lines. 7 Note that the development of cellular phones decreased the use of ground lines in France. This could potentially impact the result. We will ignore that problem in this paper. 8 Tables 2b and 2c in the Appendix show the market shares of the different operators in France.
6
2000), and 72% of the CAC 40 (the 40 most important companies in the French stock market) are customers of Cegetel (Cegetel data from www.cegetel.fr, accessed on 25th of September 2005). The dataset also reveals a surprising trait: older agents know less about the market and are less likely to switch (less than 2% of them were clients of another operator in 2000). One explanation could be that they call less often than they are called and this would offset the time they have to consider the different options. Moreover, the wealthier the agent, the better she knows the telecommunications market: 33% of individuals with revenues inferior to $1200 per month do not know any alternative to France Telecom in 20019. Also the more equipped in telecommunication devices a household is, the more likely it will switch. Hence it seems to be more difficult for individuals to switch than for bigger agents. This highlights the importance of reconsidering some economic behavior through defaults: we should analyze the costs attached to a switching decision, not only economically but in terms of procrastination costs. Defaults clearly affect the economic behavior of individuals in telecommunications (same phenomenon for cell phone, where individuals do not usually switch): but this can be generalized to different aspects of the lives of individuals: people would not choose to change companies and prefer staying in a default company and get a job perhaps inferior to a job they could have obtained had they decided to switch companies, choose to keep the same operating system instead of choosing another one less prone to viruses (Microsoft by default), use the same internet browser (Internet Explorer instead of Netscape) and buy the same brand of car year after year. Companies and governments make it even harder to switch and reinforce the default: creation of fidelity plans and rebates the longer consumers stay with the same company, and waiving some fees (banks today provide free checking has with direct deposit that secure the paycheck of clients for the bank) for example. In the next section we will focus on cadaveric organ donations, considering the legislation as the default rule10, i.e. is legislation the reason why we see differences in organ donations between countries?
9
Table 2d in the Appendix shows the Telecom Budget of French Household according to their revenue. We see that consumption of telecommunications increases with revenue. 10 Using cadaveric organ donations we can consider that the incentives and direct benefits from an individual being placed in a default will not alter his decision to donate organs (as increasing the amount of savings later in life, decreasing the phone bills by switching operators, changing job to increase one’s wage would be a bypass product of the default. Here the final result is having more organs for patients.
7
III. Defaults and Cadaveric Organ Donations11 We want to see the main consequences for a country to choose presumed consent (donor by default) or informed consent (non-donor by default) in terms of cadaveric organ donations. Presumed consent is usually perceived as the better system in terms of donations, as it will generate more possible organ donations, and as it seems that more organs can be harvested under presumed consent. In this paper we will show that the informed consent system could actually turn out to be the more successful one if the presence of the family in the organ donation process is taken into account.
1. The “Common Sense” Model Without Family
We detail in this first part the common way of tackling the impact of legislation on cadaveric organ donations. We consider a model of organ donations without taking into account the decision of the family. We will denote by u d the utility of an individual if she donates her organs and u k if she decides to keep them. v = u d − u k is the difference between these utilities and hence denotes the value of donation for the individual: if v f 0 , she wants to donate her organs, whereas if v p 0 she wants to keep them. Let c be the cost of donating organs in IC countries and the cost of keeping them in PC countries12. Registration to donate organs or not is a very simple and straightforward procedure (sending a letter, an email or make a free phone call). As a consequence, the cost c in the model is not a standard cost. It reflects the tendency of individuals to procrastinate (Akerlof (1991), Madrian and Shea (2001), Thaler and Benartzi (2004)) or the moral cost of thinking about your own death (National Health and Medical Research Council (1997)). The existence of this cost is justified in the medical literature, as individuals want to avoid thinking about their death or follow the procedure to donate organs. Gallup (1993) showed that 85% of the respondents supported organ donations, but only 37% were very likely to donate their own organs, and 25% were not at all likely. Moreover, 36% of the respondents answered that thinking about their own death makes them uncomfortable.
11
The following model is similar to Fevrier and Gay (2004). We do not distinguish different costs for each system, as it would basically result in the same conclusions, but make the intuition more difficult to follow.
12
8
Figure 1 describes both the IC and PC models. In the IC case, she keeps her organs by default but can take the decision to donate them for a cost c.
Figure 1: Informed Consent and Presumed Consent Without Family Decision
Proposition 1. There are more organ donations in PC systems than in IC systems.13
Proposition 1 shows that in order to maximize the number of organ donations, in the case without families, a government would want to adopt the PC system. The difference between the two systems comes from the fraction of people who would stay in the default system irrespective of the system. In a PC system, these individuals give their organs “by default” whereas they will keep them under the IC system. Like in the savings case, defaults impact only individuals who did not take an active decision on the default (opting out of the default or deciding to stay in it). Had giving her organs been important for an individual who chose not to, v would have been greater than the cost c and this individual would have registered in an IC system to give her organs. Our analysis and conclusions hence hinge on the fringe of individuals that are between 0 and c in the IC system.
Moreover not taking an active decision regarding organ donations could result in an unethical effect of the default, as proposition 2 shows:
Proposition 2. None of the two systems fully respects the will of the people who do not register.
13
Proofs of the different propositions will be provided at the end of the paper.
9
Under PC rules, society may take the organs of some people against their “will”. Under IC rules, society may not take the organs of some individuals that would like to give them.
In practice, the debate on organ donation relies only on the two propositions above but in a truncated way, as IC is usually considered more ethical: in January 2004 the British Parliament debated a switch in their organ donation policy in “The Organ Donation (Presumed Consent and Safeguards) Bill”. The advocates of the PC system argued that the number of organ donations increases in such a system, which is beneficial to society because of the shortage of organs. On the contrary, the proponents of the IC system use ethical reasons for arguing that it is unacceptable to take the organs of a person if she does not explicitly agree14. Surprisingly, the impact of the family is never mentioned in the usual discussion for the choice between presumed consent and informed consent system. We will show in the next section that the presence of family can change the way we look at both systems.
2. The “True” Model: Impact of the Family
The previous framework does not consider the impact of the family, which is very important in practice (May et al. (2000)). Gallup (1993) shows that although 42% of the respondent made a decision on organ donations for themselves, 75% of them were undecided when it came to take a decision on a donation of a family member’s organs. The presence of the family could hence potentially impact the agents that did not take any decision in the previous model: under the PC system (respectively the IC system) we cannot distinguish an individual that has not thought about organ donations when alive or has procrastinated ( − c p v p c ) and whose will is unknown, from an individual that really wants to (resp. not to) donate her organs ( c p v (resp. v p −c )). We will show that our results will change under the presence of the family and that, surprisingly, the informed consent legislation should result in more organs donations.
We suppose that the family has a utility u df if they donate the organs of the deceased and a utility
u kf if not. We will denote by v f = u df − u kf the difference and we suppose that v and v f are
14
They however never mention that the IC system does not either respect the “will” of some people that would like to donate their organs.
10
common knowledge. We finally suppose in this part that there is no cost to the family to take its decision.
Figure 2 represents the game with the family. In both models, if the individual decides to register, the family has no choice but to respect her decision. If, on the contrary, the individual stays in the “default” situation, the family takes the final decision15.
Proposition 3. If the family has the same preferences as the individual, both systems are equivalent. If the family can have different preferences, there are more organ donations in the IC system than in the PC system.
The first result is quite intuitive. If the family and the individual always agree, individuals do not register and let the family take the right decision when deceased. This, however, is not what we observe as people do register in both systems.16 Many reasons can explain that the value of the family differs from the value of the individual. The family can have their own preferences and take their decision by considering both their preferences and the preferences of the individual. Doctors also complain about people not speaking enough about this issue when alive and that the family has no idea of the stance of the individual on organ donations (Rocheleau (2001)). In that case, one can suppose that family members take their decision using their own valuation. The difference between the individual will and the family’s opinion is a real concern. Recently, the British Parliament passed a Bill on that subject and UK Transplant chief executive Sue Sutherland said: “The Human Tissue Bill reinforces the importance of consent for donation but makes it explicit that it is the wishes of the individual that should prevail. Many people have found it difficult to accept that relatives can overturn those wishes and this Bill deals with those concerns”.
15
This model is conformed for example to the decision process in the Netherlands (IC) and in Belgium (PC). In most countries however, in both systems, the family can decide not to respect the will of the deceased, even if he/she registered when alive. This event is however rare. One can extend the model to let the family decide even when the deceased had registered. If the family has a huge cost to change the decision of the deceased when expressed, the conclusions will be very similar to the simpler model presented here. 16 In 2003 for example, about 37% of the Dutch population have made a registration to donate their organs (The Netherlands is an IC country).
11
Figure 2: Informed Consent and Presumed Consent With Family Decision
When taking into account the family decision, we surprisingly find the opposite of the usual claim. The IC system leads to more organ donations than the PC system, because the argument of the model without the family does not apply anymore. When an individual does not want to think about her death or procrastinates ( − c p v p c ), she lets her family decide for him/her. Both systems are thus equivalent for these agents. However, the fact that the family can take the decision for the individual, even when she has clear preferences concerning organ donations ( v p −c or v f c ), has a perverse effect when the family and the individual disagree. Registering is a protection against the family and a commitment towards organ donations. In the IC system, the individual can register and commit to give her organs. In the PC system, on the contrary, the individual can only register to commit to keep her organs. This commitment effect explains why the IC system potentially leads to more organ donations. In practice, most presumed consent countries seem aware of this commitment problem and offer the possibility to register as a donor in the PC system. Kluge (1997), for example, asserts that the shortage of organ donations would decrease if the donor card is used as a proof of donation to bypass some families reluctant to donating. This donor card only exists to try to protect the individual against her family.17
Corollary 1. Introducing a donor card in the PC system helps an individual to give her organs when the family does not agree. There are still more donations in the IC system than in the PC system with donor cards. 17
Austria has a very strict PC system where the decision of the family is never taken into account and individuals who do not register will be donors. In this country, no donor card exits.
12
One can argue that both systems are quite equivalent because an individual can commit him/herself by writing a note saying that she does or does not want to give her organs. Nonetheless this note will presumably be in the possession of the family and this supposes that the family would respect the will of the deceased. It is more difficult to credibly commit with a note than by going through an official process. To restore the equivalence between the two systems, it would be necessary to introduce a non-donor card in the IC system. Generally such a non-donor card does not exist. Nonetheless, the Netherlands have started to implement a full information registration concerning organ donations: individuals register in the National Organ Donation Register and decide if they want or do not want to donate their organs. Such a system eliminates the uncertainty of people not registering and deciding that they did not want to give their organ. Nonetheless people still need to register to enter their choice and only 28% of the population did that in the last 4 years.
The commitment problem faced by the agents when the family has a say leads to another surprising result when looking at ethical and moral reasons.
Proposition 4. If the family has the same preferences as the individual, both systems respect the will of all agents. If the family can have different preferences, none of the two systems respects entirely the will of the people. The PC system with donor cards is however more conformed to the will of the individuals than the IC system.
This result is quite surprising for several reasons. First, introducing the family seems worse in terms of ethical and moral reasons than not asking for their decision as soon as some disagreements exist between the family and the deceased (whereas doctors and ethicists claim that introducing the family say enhances the moral of the process). The will of individuals, who have clear preferences towards organ donations, is not respected anymore because of the commitment problem they face when disagreeing with their families. Second, the proponents of the IC system are usually deeply concerned with taking the organs of someone against her will. Our result proves that, actually, this event will happen more frequently in an IC system than in a PC system and will strictly affect people who would have taken a binding decision not to donate their organ donations if they had this opportunity. Here, however, the family, and not society, is responsible for not implementing the choice of the deceased. People usually consider that this is not as ethically and morally unacceptable as when society decides.
13
Third, introducing a donor card is very important on ethical and moral grounds as it helps respecting the will of the deceased. May et al. (2000) argue that, even if the family does not consent to donation, respecting the documented wishes of a deceased to donate is not only morally permissible, but morally required. Chouchau and Draper (2003) also argue about the superiority of the PC system with a donor card when it comes to the respect of the will of people.
When ethical and moral reasons are debated, it seems that politicians, doctors, and religious people, all agree that the right thing to do is to respect the will of the deceased. However, this may be questionable. A standard hypothesis in economics is that the utility of deceased people is zero. As a consequence, the only utility that should matter in the end is the families’. In such a case, none of the legislations is ethically and morally perfect. The “best” system would be a system in which the individual has no say and in which the family has all the power to take a decision regarding organ donations. It is hard enough that a family member passed away. Perhaps, the family should not have also to bear on top of that the fact that they must respect the will of the deceased if her decision is painful for them. In a sense, most countries maximize the utility of the living by giving them the final decision. Still, to reach that goal, the commitment effect that exists in both systems should be eliminated.
3. General Case: Procrastination and Death Taboo
The previous models (with and without family) are two extreme cases of a more general model where the family is also affected by the default situation. If the family bears a cost c f of not staying in the default situation, the model without family (Part III 1) corresponds to c f = +∞ whereas the model with family (Part III 2) corresponds to c f = 0 . In the general case, the effects derived in the extreme cases are both at stake. Two opposite forces are actually conflicting and each one could potentially be dominant. Figure 3 represents the equilibrium in the general case. Both systems are equivalent except in three regions. In the middle region, taking a decision is too costly for both the individual and the family. The choice directly comes from the default situation.
In such a case, the PC system dominates the IC system. In the two other regions, the individual and the family want to make a decision, but disagree. The commitment effect is central here, and
14
the IC system dominates the PC system. The total effect is ambiguous and depends on the values
v and v f and on the costs c and c f . The choice of an “optimal” system in a given country should thus depend on the behavior of people regarding their death or the death of a relative.
Figure 3: Comparison of Both Systems in the General Case with Families
The other explanation expressed by doctors to understand why people do not register is called death taboo (National Health and Medical Research Council (1997)). Unfortunately, it is difficult to have an objective measure of death taboo.
A good proxy can be the willingness of people to let the family know or not about their post mortem decisions. In Quebec for instance, only 50% of the population writes a testament18. In 4 cases out of 5 this testament is registered by a notary. Moreover these testaments usually deal with the succession but very few people let specific instructions regarding their funeral. These figures underline the importance of death taboo. People do not like to think about their death and to what will happen to their body once deceased. This argument also emphasizes that the cost for the individual may be important19. 18
These figures were given to us by the Chambre des Notaires du Quebec. This cost may vary by individual. The richer, the older and the more educated an individual is, the more likely s/he is to write a testament and therefore to think ahead about their death.
19
15
The family, contrary to the individual, has no choice but to think about the situation and take a decision quickly after the death of their relative. They have to decide once and for all in a short amount of time and cannot postpone this decision. Ignorance, procrastination and death taboo thus appear to be less relevant for the family which is expected to have a smaller cost c f than the cost c to the individual. Nevertheless, this cost may not be zero. The family has a difficult decision to take and faces anticipated regret. It is thus possible that the family procrastinates in the sense that they do not want to take such an unpleasant decision. Moreover, in PC countries, by refusing to donate the organs, the family kind of “kills” someone (a potential recipient) who should have lived under the strict application of the law. Such a decision, under the pressure of medical doctors, may be hard to take. In IC countries, on the contrary, members of the family have the law on their side when deciding not to give organs. If such effects are at stake, the cost for the family may not be negligible. Given the expected structure of the costs (high cost for the individual and smaller cost for the family), one crucial element is the distribution of the preferences (v, v f ) . If the individual and the family have well correlated preferences, the PC system may be a better one. As already mentioned, this is not necessarily the case. Because of the death taboo, the family often has no idea of the will of the deceased. This absence of communication on organ donations can lead to more disagreements between the choice of both the individual and the family. For these reasons, when death taboo is an important issue, it gives some arguments for a country towards choosing the IC system. Furthermore, the family is known to be much more reluctant than the deceased to tackling the issue of organ donations. It is thus possible that in a PC country, an important number of people would like to give their organs but cannot commit themselves against their family which leads to the conclusion than the IC system could be better in terms of organ donations.
These results show that it is actually more difficult to choose between both defaults. We also show that the advocated superiority of presumed consent systems is not such a foregone conclusion. In the next part we consider the example of Sweden that switched between both systems twice and show that the evolution of organ donations is different from the expected results.
16
IV. Sweden: No Effect of Switching Defaults Sweden changed its legislation towards organ donations twice in the last 30 years: until 1986, Sweden was a presumed consent country, where the family had a say in the decision for deceased individuals that did not register for donations while alive. Between 1986 and 1996, the country switched to informed consent, where a donor needed to register while alive to donate her organs once deceased20. Finally, Sweden came back to the presumed consent system at the end of 1996. The goal of the last switch was at the time to spark donations, as they were steadily decreasing since the beginning of the 1990’s.
We obtained figures for organ donations in Sweden between 1981 and 2004 from Hakan Gabel of the Swedish Department of Health (SOS). We graph in Figure 4 the evolution of organ donations per million population over the period 1981-2004. We can distinguish five main periods: low donations during the first 3 years of data reporting 1981-1983 (around 13 pmp), a stable period of high donations 1984-1989 (donations between 16 and 18 pmp), the ongoing decline in donations over 1990-1996 (donations decreased from 15 pmp to 11 pmp), the “information” period 19961998 due to the new register, and the final period 1999-2004 (donations have been under 14 pmp) - even if since 2002 donations have increased from 11 pmp to close to 14 pmp. We also reported on the graph the different periods when presumed/informed consent system was in place. Overall, between 1981 and 1986, under the PC regime, there was an increase of 37% of organ donations (even if this figure should be taken cautiously as there might have been some reporting issues for the first few years the donations were counted). Under the IC system (1986-1996) organ donations decreased by 38%. Over the period of presumed consent 1996-2004 the increase was limited overall (8%) but the trend of the last two years seems on the mend (a rise of 25% between 2002 and 2004).
The decrease in organ donations during the IC legislation seems to corroborate the usual idea that IC systems do not provide enough organs compared to PC legislations. Nonetheless the measure corresponds to “per million population” and not necessarily to the population of reference when it comes to donations, like the deaths of individuals between 18 and 75 years old (that would correspond to the maximum potential cadaveric donors at the doctors disposal as there are almost no donors younger than 18 years old and older than 75) and motor-vehicle deaths (MVA) deaths
20
Politicians invoked at the time ethical issues as the main reason for the switch.
17
(usually an important part of cadaveric donations when organs are harvestable and individuals gave the permission to take their organs). Evolution of Organ Donations in Sweden Over the Period 1981-2004
19
18 PRESUMED CONSENT 17
16
15
14
13
12 PRESUMED CONSENT 11
INFORMED CONSENT
10 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Figure 4: Evolution of Organ Donations in Sweden over the Period 1981-2004
Over the years, using US data (UNOS), we notice that the share of organ donations coming from motor-vehicle related deaths was constant and around 20-25% over the period 1990-2002. This is the most stable part of the organ donations throughout the years in the United States; hence we have a more reliable measure of the overall donations if we consider the ratio of organ donations to fatal car accidents and if we suppose that this ratio still applies to the Swedish case. Therefore the choice of the measure of reference is pivotal in the study of the impact of legislation in organ donation, given that legislation should provide more organs for a given population of potential donors (because of their characteristics) and not the entire population (the usual measure in the literature is the per million population). Indeed, when using this new measure we distinguish a different trend between both ratios: the ratio using the population leads to a decrease in organ donations, whereas the ratio on the MVA presents an increasing trend.
We show in Figure 5 the evolution of MVA deaths in Sweden over the period 1991-2004. There was a clear decrease in the MVA deaths between 1989 and 1996, from 105 to 60 deaths pmp.
18
Moreover the variation of MVA deaths vis-à-vis organ donations is an important component of the organ donation pattern: when comparing fatal car accidents data from the World Health Organization Mortality Database with the cadaveric organ donations data, we found that MVA deaths decreased at the same time as organ donations were declining.
110 105
PRESUMED CONSENT
100 95 90 85 80
INFORMED CONSENT
75 PRESUMED CONSENT 70 65 60 55 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Year
Figure 5: Evolution of the MVA Deaths in Sweden in per million population
Moreover there is an overall important decrease of the number of fatal car accidents between 1981 and 2004 (-25.6%), whereas organ donations actually increased (10.5%) over the same period. Hence organ donations potentially increased compared to the pool of potential organ donors.
We clearly infer from this trend that organ donations have actually increased if we consider the ratio of organ donation to fatal car accidents that represent the main cause of eligible donations. Measure for Organ Donations Per Million Population Per 10,000 Deaths Per 100 MVA Deaths
PC Until 1986 15.16 13.76 15.85
IC Between 1987 and 1996 14.56 13.35 17.82
Table 3: Change in Swedish Organ Donations Using Different Measures
19
PC After 1996 12.44 11.83 18.22
Table 3 shows that there was a steady decrease of the number of organ donations throughout the period 1981-2002 considering the per million population measure. On the contrary, the measure per MVA deaths underlines an increasing number of organ donors with respect to the population of reference. These two estimates do not give any clear relationship between legislation and organ donations. The donations per 10,000 Deaths of the potential organ donors (deaths per year between 15 and 75 years old) corresponds to the maximum population of reference for organ donations and is taken as a benchmark.
Considering the per million population measure, we have a steady decrease in the number of organs received, irrespective of the legislation. It seems that the cadaveric donations in Sweden went down between the entire period between 1986 and 2004 considering the per million population or the per 10,000 deaths measure, whereas the MVA population of reference gives interesting results recouping the theory developed before: the evolution of organ donation has followed the quick increase predicted by the model (+2.97 per 100 MVA over the entire period, and +2 between 1986 and 1987). The increase in 1997 (hence the return to presumed consent) is quite noticeable when using the per 100 MVA measure (+2.27 between 1996 and 1997 and lower increase +0.40 over the entire period): this is the result of the information effect where promotion of the new legislation and advertisement in favor of donations are made. Promotion is often disregarded as an influent vector of organ donations. Nonetheless, advertisement campaigns figuring stars have usually resulted in more donations, irrespective of the legislation (Michael Jordan as a spokesman for donations in the US in 1996 for instance21).
Figure 6 gives the evolution of organ donations during the different time periods using the different measures for organ donations. The blue line corresponds to the evolution of per million population donations. We see that under presumed consent cadaveric organ donations were between 13 and 18 pmp. Then during the period of informed consent legislation there was a slow decrease (as of 1990) from 18 pmp to less than 12 pmp. There was then a short recovery for the first two years (1996-1998) under presumed consent but donations stayed under 14 pmp (1993 was the last year to have donations over 14 pmp).
The peak in 1996 seems to highlight the importance of the new register in Sweden. The register is a combined registered where non donors are usually overrepresented: 30% of the registered individuals agreed to donations. 21
http://www.transweb.org/news/press/archive/pr_jordan_4_17_96.html
20
These results show that it is actually more difficult to choose between both defaults. We also show that the superiority of presumed consent systems suggested by politicians is not such a foregone conclusion.
Organ Donations in Sweden With Different Measures 26
24
Presumed Consent
Informed Consent
Presumed Consent
22
20
18
16
14
12
19 81 19 82 19 83 19 84 19 85 19 86 19 87 19 88 19 89 19 90 19 91 19 92 19 93 19 94 19 95 19 96 19 97 19 98 19 99 20 00 20 01 20 02 20 03 20 04
10
Donations PMP Linear (Donations per 100 Car Accidents)
Donations per 10,000 deaths Linear (Donations per 10,000 deaths)
Donations per 100 Car Accidents
Figure 6: Evolution of Organ Donations in Sweden with Different Measures for Donations
Nonetheless the Sweden case does only gives a glimpse of answer concerning change in legislation: we could advocate that the change was PC-IC-PC and perhaps IC-PC-IC would have drawn different conclusions, as a country being in a default informed consent might have done more towards promoting organ donations (as the default is “not donating”). This is why a more thorough empirical study should be made on legislation and organ donations. For that matter, in the next section, we consider the impact of legislation on a panel data of countries between 1990 and 2002.
21
V. Empirical Estimation of the Model We have seen in Figure 1 that informed consent (IC) countries lag behind presumed consent (PC) countries in terms of raw donations per million inhabitants. This is the reason why policymakers in IC countries usually try to switch to PC in order to increase donations. Nonetheless we showed in our model in Part III that IC should in theory result in more donations. In this section we corroborate the results of our model using countries with official registers, i.e. where preferences of the individuals are known to be registered 22 . We want to measure the importance of the default legislation on organ donations in such countries. The countries with official registers are: Australia, Austria, Belgium, Canada, Denmark, France, Hungary, Ireland, Italy, The Netherlands, Poland, Portugal, Sweden, UK and the USA23. Details on the registers are given in Table 4.
We will estimate the impact of legislation on organ donations considering the different measures of organ donations and testing if the legislation has actually a sizable effect on donations considering the presence of the family.
We consider the following model:
Yit = α Dit + β X it + ε it .
Yit is the logarithm of the organ donation rate in country i at time t (it can represent donations per million population or donations per hundred MVTA potential donors or donations per 10,000 reference deaths i.e. deaths of individuals between 20 and 75 years old). Dit is a discrete variable indicating the type of organ donation law implemented in country i at time t (presumed consent legislation corresponds to the 1 value). X it represents the matrix of covariates (GDP, health
22
Some countries indeed do not use a registration system that enforces the decision of the individual (i.e. even if a donor wants to register not to give her organs once deceased for example, no register exists and doctors will ALWAYS ask the family before harvesting organs). The countries we focus on have an official “yes” or “no” register during the time period 1993-2000. These countries represent an adequate dataset to test our model, as once the individual opted out of the default the family cannot change her will. 23 In the US, 32 states out of 52 have an official register.
22
expenditures, Catholic, and MVA&CVD deaths 24 ). All specifications will include year fixedeffects and errors will be clustered by country.
The dataset comes from Abadie and Gay (2004). The database contains organ donation information, legislation, GDP, and other demographics for these countries, over the period 19902002. International data on organ donations come from the Transplant Procurement Management (TPM, 2003) and Organización Nacional de Transplantes (ONT, 2003). We only consider kidney donations in our paper. Organ donations are usually measured in per-million-population units (pmp), i.e. the number of donations divided by the population, but we will also consider the measure of reference of per-hundred-potential MVA-CDV deaths potential donors defined earlier and the per 10,000 20-75 year old deaths reference deaths. The data on legislation on organ donations was compiled by consulting the legal literature or by direct contact with organ donation agencies for each country. The different legislations for all the countries in our sample are detailed in Appendix. GDP data comes from the World Bank Development Indicators (World Bank, 2003a). The number of deaths caused by traffic accidents and cerebro-vascular failures were taken from the World Health Organization Mortality Database (World Health Organization, 2003). Data on religious beliefs and legal systems come from CIA (2003). Blood donations are taken from the International Federation of Blood Donors Organizations (FIODS). Blood donations will correspond to a measure of the willingness to give from the donors in each country, as some countries might be intrinsically more inclined to give than others. Controlling for such a subjective measure is difficult as donation of organs differs from blood donations (as blood donation does not necessitate thinking about death or does not involve any removal of any parts of one’s body) but we did not find a better measure of the willingness to give. Another idea could have been monetary donations in the Telethon for different sicknesses, but it would have been harder to control precisely for media exposure in favor of donation in each countries.
Sample statistics are presented in Table 5 in the Appendix. In our sample presumed consent countries have slightly higher cadaveric donations than informed consent countries (15.43 against 15.26). The GDP per capita is higher for informed consent countries, as are health expenditures. Most of the presumed consent countries are Catholic contrary to informed consent countries (0.919 versus 0.202). Moreover presumed consent countries do not have common laws, compared 24
These variables are the usual determinants of organ donations (see Abadie and Gay (2004) for more details). GDP, health expenditures and MVA & CVD deaths are considered per capita in all of our regressions. Using health expenditures did not change any of the results presented here, as it is a percentage of GDP. I will not report these results here.
23
to informed consent countries (0.262). Last, motor-vehicle accidents and cerebral vascular diseases are more important in PC than IC countries (1.546 per ten thousand population against 1.197).
Tables 6a, 6b and 6c give the results of the regression distinguishing between the three different measures of organ donations detailed in Part IV for the Swedish case. Table 6a uses the conventional per million population measure of organ donations. Legislation has no significant effect on organ donations without controlling for other determinants. This result corroborates the fact that legislations in our model can be indistinguishable in some areas of Figure 3 in Part IV. Nonetheless, presumed consent legislation has a significant negative effect (-36%) on organ donations, after controlling for GDP, religious preferences, MVA deaths and blood donations (column (6) of Table 6a). We tested the robustness of the results in columns (2) to (5) of Table 6a. We see that the negative effect of presumed consent legislation is robust and the values vary between -25% to -36%. Tables 6b and 6c detail the same regression with a different measure of reference for cadaveric donations. Table 6b focuses on the number of deaths between 20 and 75 years old. We find a negative25 effect of the presumed consent legislation on organ donations (-36%) once controlling for the same determinants. The variables considered in this regression seem to explain more the variance of the dependent variable ( R 2 = 0.634 instead of R 2 = 0.509 ). We should observe that standard errors are smaller in this case than in the “per million population” one). Table 6c uses the MVA deaths as the population of reference for measuring organ donations. We find a significant negative effect of presumed consent system, even higher than the last two cases (-44%) due to the higher decrease of MVA deaths in IC countries. These results are somewhat robust and vary between -30% and -44%. We see that IC countries have an important effect on organ donations without controlling for any other determinants as MVA deaths are smaller in IC than PC countries.
In conclusion, we find a significant positive effect of the informed consent system, with some level of robustness, irrespective of the measure of the cadaveric organ donations. Indeed the impact of presumed consent legislation is found to be negative (between -26% to -44% depending on the measure of organ donation considered) in countries with a register in place, when controlling for GDP, religious beliefs and the log of MVA deaths. Legislation seems to have
25
This is of similar order as the per million population case in Table 6a.
24
some direct effect on organ donations in countries with registries. A reason could be that these countries are placed in the region of Figure 3 in Part 3 were IC is superior to PC because of costs considerations.
Nonetheless, we thought that there could be a potential simultaneity between organ donation and legislation. In order to fully identify the effect of legislation on organ donation, we need an instrumental variable that affects the presence of presumed consent legislation and that does not belong directly to the production function of donating organs. We consider common law as our instrument in the estimation of our equation.
If this common law variable is a good instrument, it will be uncorrelated with the donations, except through variables included in the equation estimating organ donation. We do not find a significant impact of common law legislation on organ donation per se. The link between informed consent legislation and common law systems stems from the Commonwealth history. English dominated countries have adopted similar political and civil system. We detail in Table 7 the results of the instrumental variable regression on the sample. The negative effect of the presumed consent legislation seems to be reinforced in the IV estimation, as we found a significant negative effect of presumed consent legislation on organ donations (between -29% and -70% depending on the measure of organ donations and the presence or not of blood donations), after controlling for GDP, religion preferences, and car accidents. This attempt to try to separate the effects of legislation on donation seems to nonetheless show the difficulty to stress in practice the impact of the legislation on organ donations. The estimate is rather important but does only concern countries with a register, which could induce a bias towards individuals who would make a decision on organ donation (especially considering countries with complete information registration as the Netherlands). The difficulty to find good instruments for presumed consent legislation brings a small weakness to our analytical test of the theory.
Actually the important point in this analysis was to show the possibility of a negative effect of presumed consent on organ donations on countries fitting the model in Part III. When considering other countries and a broader application of the family role, presumed consent countries seems to have a positive impact on organ donations (see for example Abadie and Gay (2004)). In the next section we will discuss the results we found in this section and give recommendations in terms of organ donations policies.
25
V. Implications of the Estimated Effects of the Impact of Legislation on Cadaveric Organ Donations, Discussion and Extensions Having derived estimates of the impact of the default legislation on cadaveric organ donations, we would like to stress the public policy implications of these estimates. Moreover, we try to consider some adjustments to our previous results, like estimating the impact of legislation once adjusting for family refusal.
First, our coefficients suggest that informed consent laws may actually increase the supply of cadaveric organs for transplantation. Nonetheless, the estimation is only made on countries where a register is in place for registering individuals’ decisions. This can be due to the sample at hand and the presence of higher than expected rates of donations in informed consent countries. Note also that Spain, the country with the highest number of cadaveric organ donations is absent from the sample. We redid the same estimation with Spain and the different estimations and the IV estimation stays significant but the coefficient decreased to reach (-24%). Note that this is just a verification that reintegrating the country with the highest number of organ donors does not completely change the results.
Second, as in Abadie and Gay (2004), the estimation on the panel data has some disadvantages: the impact of the legislation is estimated up to the unobserved heterogeneity between countries that we did not manage to take into account. Indeed donations differences, even within countries. It is almost impossible to take into such differences and try to explain the different preferences for donations for the different countries. One important criticism of the data is the significant difference in GDP between informed consent and presumed consent systems ($4,782 per capita) that might bias the results towards a negative effect of PC on organ donations. We tried to evaluate the effect of this GDP by taking the countries with the highest and smallest GDP in each group (IC and PC) and found that the result was still negative and significant.
Third, we should take into account the decision of the families in our estimations in terms of potential organs that are lost. We report in Table 8 in the Appendix the available percentages of family refusal in Spain, France, Italy and Australia between 1996 and 2002. Usually refusal rates by families in IC countries are higher than in PC countries. This could result from the influence of the defaults on the family’s decision itself. Taking these measures of refusal by families into account, one could measure the effect of presumed consent and informed consent on organ
26
donations. We take these figures into account imputing the organs lost because of family refusal (higher in IC countries than in PC countries). We find a stronger negative effect of informed consent on organ donations. There are other possible methods to reintegrate the impact of family refusal in the regression, but the small amount of data available does not make inferences on refusal decisions possible.
Fourth, in order to reduce the impact of the defaults on people’s decisions and maximize ethical issues, countries should resort to double registries like in the Netherlands, an informed consent country, which has the particularity of having a double registry. Once registered, the wishes of the individuals are known, and it is impossible for the family to object to it. Nonetheless, only 37% of the Dutch registered so far. For the other 63%, doctors have to ask the family relatives to make a decision about donation. The family refusal is at the moment around 80% for 2003, but the remaining 20% are donors and families cannot change the person’s decision (Verzijden and Schothorst (2003)).
Furthermore, we could think of institution a price for donation, breaking the altruistic system that showed its limits: Becker and Elias (2003) show how to suppress the gap between supply and demand of organs by instituting a market and an equilibrium price for living donors. They evaluate the value of an organ, considering forgone earnings, decrease of quality of life. Their estimate is between $20,000 and $45,000. Other authors have found that monetary incentives would help solve the lack of supply of organs: Kaserman and Barnett (2002) estimate that the organ shortage would be resolved at prices between $1,000 and $3,000 per living donor. Applying the same idea in the case of cadaveric donations, an attempt has been made by the State of Pennsylvania to pay $300 toward the funeral expenses of the deceased if families agree to donate organs. It was never implemented because of the conflict with the Uniform Anatomical Gift Act of 1987. Nonetheless the creation of a small incentive ($10-$50) would certainly make people register, in order to compensate individuals for lost time. This would increase the probability of registration (only 28% hold a signed ID for organ donation in the United States, as over the years attitudes towards organ donations have evolved positively: the Gallup survey on organ donations (Gallup, 1993) indicates that most Americans favor organ donation (85%), and would like to donate their organs after death (69%).
Last, the main point of this paper is to increase the awareness that legislation, contrary to the usual debate, is not the solution to the shortage of organ donations. The real solution might be
27
increasing the awareness that organ donation saves lives: throughout this research, I found that peaks in each countries are recorded when registries are put into place or when organ donor awareness campaigns are launched. More research is needed on information to individuals concerning organ donations. The lack of knowledge is prevalent in the case of organ donations:26: in France, so few people knew that the country was presumed consent that a law was voted in 1996 to reinforce presumed consent knowledge, although it existed since 1976! Moreover, defaults applied to cadaveric organ donation also have a major problem: they do not actively require the individual to take a decision and that involves other people’s needs, not only their own27. I compared such an organ donation phenomenon as the different telethon organized to subsidize medical research against sicknesses. They were created to inform individuals about the existence of how a particular disease (not necessarily know by everybody) affects other people’s lives through a 36 hour- television program. Table 9 reports the values of the money raised by the French Telethon. Exposure of the sickness on television has brought an increasing level of donations, which could serve as a benchmark for measuring the potential for awareness on organ donations.
VI. Conclusion This paper analyzes an important (but heretofore ignored in economic literature) problem of stressing the impact of defaults on economic behavior of agents. We showed that, in the organ donations process, families twist the default into turning a presumed consent system (that would ensure more organ donations at first sight) into a worse system than informed consent legislation. Some medical studies (usually written by doctors from informed consent countries) have pointed out that, on average, presumed consent countries do not produce significantly higher organ donation rates. Moreover, several authors have hypothesized that this lack of correlation is produced by the fact that presumed consent laws are rarely enforced and that, in practice, family consent is always required before organs are extracted. This paper reinforces that trait and shows that the presumed consent default actually makes such a result possible: families exploit the fact 26
Such a lack of knowledge of the market also exists in the telecommunication example in Part II. Table 10 in the Appendix presents how customers are unaware of the different possible offers and stick to the default provider. 27 This is why organ donations are a key example for testing the impact of defaults on individual behavior. In the savings case, increasing/ opting-out of a default rule has a clear effect of the future income of the concerned individual.
28
that they can overrule the default and they decide not to give organs of deceased that did not register and nonetheless wanted to give their organs.
Pushing the analysis further we might discover that the default should not matter when full information is revealed. The Swedish example shows that switching legislation does not necessarily increase donations, and that better information or promotion campaigns have more impact on individuals, making them responsible for the death of patients that could be saved if they agree to donate their organs once deceased.
We reinforced this analysis by showing that informed consent legislations actually result in more donations than presumed consent legislation on a dataset consisting of countries with existing official registries. Such a tradeoff would disappear under a full information registry, as in the Netherlands, where people communicate their exact preference (donation or non donation) at the moment of registration. More research is needed to understand the impact of information on defaults to fully understand how defaults affect economic behavior.
29
References Abadie A and S. Gay. 2004. “The Impact of Presumed Consent Legislation on Cadaveric Organ Donation: A Cross Country Study”, NBER Working Papers 10604, National Bureau of Economic Research. Akerlof, G.A. 1991. “Procrastination and Obedience”, American Economic Review, 81(2), 1-19. ANZOD. 2003. The Australia and New Zealand Organ Donation Registry. 2002 Annual Report. http://www.anzdata.org.au/ANZOD/ANZODReport/anzodreport.htm ART. 1998. Authority of Regulation of Telecommunications. Survey on Consumers Attitudes towards Deregulation of Telecommunications in France. http://www.art-telecom.fr/ ART. 2001. Authority of Regulation of Telecommunications. Attitudes of Small Governments on the Change in Telecommunication Providers in France. http://www.art-telecom.fr/ ART. 2004. Authority of Regulation of Telecommunications. State of the Telecommunication Sector http://www.art-telecom.fr/ Becker G.S. and J.J. Elias. 2004. “Introducing incentives in the market for live and cadaveric organ donations”, Mimeo. British Human Tissue Bill. 2004. http://www.publications.parliament.uk/pa/cm200304/cmbills/049/04049.i-iii.html. CIA. The World Factbook. 2003. http://www.cia.gov/cia/publications/factbook/. Choi J. , D. Laibson, B. Madrian and A. Metrick. 2001. “For Better or For Worse: Default Effects and 401(k) Savings Behavior”, NBER Working Papers 8651, National Bureau of Economic Research. Choi J., D. Laibson, B. Madrian and A. Metrick. 2003. “Passive Decisions and Potent Defaults”, NBER Working Paper N. w9917, National Bureau of Economic Research. Chouchan P. and H. Draper. 2003. “Modified mandated choice for organ procurement”, Journal of Medical Ethics, 29, 157-162. Consumer Reports Magazine. 2005. Cell Phones, www.consumerreports.org Fevrier P. and S. Gay. 2004. "Informed Consent versus Presumed Consent: the Role of the Family in Organ Donation", Mimeo University of Chicago. FIODS. Fédération Internationale des Organisations des Donneurs de Sang. 2004. http://www.fiods.org. Gallup Inc. 1993. "The American Public’s Attitudes toward Organ Donation and Transplantation." Conducted for The Partnership for Organ Donation, Boston. Johnson E. and D. Goldstein. 2003. “Do Defaults Save Lives?”, Science, 1338-1339.
30
Kahneman D., Knetsch J. and R. Thaler. 1991. “The Endowment Effect, Loss Aversion, and Status Quo Bias: Anomalies”, Journal of Economic Perspectives, 5(1), 193-206. Kahneman D. and A. Tversky. 1979. “Prospect Theory: An Analysis of Decision under Risk”, Econometrica , 47(2), 263-292. Kaserman, D., and A. Barnett. 2002. “Alternative Policy Proposals: A Survey and Comparative Economic Analysis”, in “The U.S. Organ Procurement System: A Prescription for Reform”, pp. 41-68. Washington, DC: AEI (American Enterprise Institute) Press. Kluge E.H. 1997. “Decisions About Organ Donation Should Rest With Potential Donors, Not Next of Kin”, Canadian Medical Association Journal, 157, 160-161. Laibson D. 1998. Psychological perspectives on 401Ks, Frontiers in the Economics of Aging, David A.Wise ed., Chicago: NBER and University of Chicago Press, 106-120. Madrian B and D. Shea. 2001. “The Power of Suggestion: Inertia in 401(k) Participation and Savings Behavior”, Quarterly Journal of Economics, 116, 1149-1525. May T., Aulisio M. and M.A. De Vita. 2000. “Patients, Families and Organ Donation: Who Should Decide?”, The Milbank Quarterly, 78(2), 323-336. Mustarah F. 1998. “Organ Procurement: Let’s Presume Consent”, Canadian Medical Association Journal, 158, 231-234. National Health and Medical Research Council. 1997. “Donating organs After Death: Ethical Issues”, Mimeo Commonwealth Department of Health and Family Services of Australia. ONT. Organización Nacional de Transplantes. 2003. Estadísticas. http://www.msc.es/profesional/trasplantes/estadisticas/estadisticas.htm. Organ Donation (Presumed Consent and Safeguards) Bill. 2004. http://www.epolitix.com/EN/Legislation/200402/0d05bd1c-d563-4014-9d15-e17d36ad8474.htm. Rocheleau C.A. 2001. “Increasing Family Consent for Organ Donation: Findings and Challenges”, Progress in Transplantation, 11(3), 194-200. Samuelson W. and R. Zeckhauser. 1988. “Status Quo Bias in Decision Making”, Journal of Risk and Uncertainty, 1, 7-59. Spital A. 1996. “Mandated Choice for Organ Donation: Time To Give It a Try”, Annals of Internal Medecine, 125, 66-69. Thaler R. and S. Benartzi. 2004. “Save More Tomorrow: Using Behavioral Economics to Increase Employee Savings”, Journal of Political Economy, 112(1), 164-187. TMP. International Registry Organ Donation Transplantation. 2003. Transplant Procurement Management. http://www.tpm.org/registry/reg_mondo.htm. UK Transplant. Statistics. 2003. http://www.uktransplant.org.uk/.
31
UNOS. United Network for Organ Sharing. 2002. "Annual Report of the U.S. Organ Procurement and Transplantation Network and the Scientific Registry of Transplant Recipients: Transplant Data 1992-2001". http://www.optn.org/data/annualReport.asp. Verzijden D. and Y. Schothorst. 2003. Orgaandonatie : alternatieve beslissystemen, eds Veldkamp, Amsterdam. World Bank. 2003a. World Development Indicators. World Bank. 2003b. Education Statistics. http://devdata.worldbank.org/edstats/cd5.asp. World Health Organization. 2003. WHO Mortality Database. http://www3.who.int/whosis/.
B. Tables Table 2a: Evolution of Base Prices for Telephone Communications in France Operators
2000
2001
2002
2003
2004
100
97.1
96.1
94.9
94.1
92.3
92.7
90.2
89.4
87.5
France Telecom Other Operators
Note: Index 100= France Telecom prices in 2000 Source: ART (2004)
Table 2b: Evolution of Market Shares for the Local Communications in France Market Share 2001 France Telecom 3 Principal Competitors Other Operators
87.2% 12.2% 0.6%
Local Communications Value Volume Calls 2002 2003 2001 2002 78.0% 21.3% 0.8%
74.9% 24.1% 0.1%
Source: ART(2004)
32
87.7% 11.7% 0.6%
76.7% 22.4% 0.9%
2003 72.5% 26.1% 1.5%
Table 2c: Evolution of Market Shares for International Communications in France
International Communications Value Volume Calls 2002 2003 2001 2002
Market Share 2001 France Telecom 3 Principal Competitors Other Operators
76.4% 19.3% 4.3%
75.7% 21.5% 2.8%
75.1% 20.9% 3.9%
71.0% 23.5% 5.5%
2003
69.7% 26.4% 3.9%
70.2% 26.7% 3.1%
Source: ART(2004)
Table 2d: Telecom Budget According to Monthly Earnings for French Households
Monthly Earnings for French Households Total
Less than 15 Euros Between 15 et 30 Euros Between 30 et 60 Euros Between 60 et 100 Euros Between 100 et 160 Euros Between 160 et 320 Euros 320 Euros and higher No Answer
Less than 900 €
900 to 1200 €
1200 to 1500 €
1500 to 2300 €
2300 to 3000 €
3000 € and more
3%
1%
2%
1%
0%
1%
25%
15%
12%
8%
4%
1%
40%
49%
41%
34%
26%
11%
21%
21%
28%
34%
32%
20%
4%
9%
13%
19%
27%
40%
2%
2%
3%
4%
8%
22%
1%
0%
0%
1%
0%
3%
4%
1%
1%
0%
3%
2%
1% 9% 33% 27% 19% 7% 1% 2%
Source: ART (2001)
Table 4: Description of the Different Countries With Registers Percentage of population/number of citizens having registered Donor Registers Combined Registers Non donor Registers Sweden (16%) Australia UK (20%) Canada ( only BC)
The Netherlands (40%) Australia (26%) Denmark (8.1%) Belgium
33
Austria (4.4%) France (0.075%) Slovakia (0.0002%) Hungary Italy Poland Portugal
Table 5: Summary Statistics (1) Entire Sample
(2) Presumed Consent Countries
(3) Informed Consent Countries
Presumed consent country
0.512 [0.501]
Cadaveric Donors (per million population)
15.356 [5.232]
15.432 [6.136]
15.269 [3.985]
GDP per capita (constant 1995 USD)
20,713 [10,214]
18,402 [11,141]
23,184 [8,501]
Health expenditures per capita (constant 1995 USD)
1,925 [971]
1,634 [946]
2,285 [880]
Catholic Country
0.597 [0.492]
0.919 [0.278]
0.262 [0.442]
Common Law
0.3155 [0.492]
0 [0]
0.553 [0.499]
MVA & CVD deaths (per ten thousand population)
1.374 [0.508]
1.546 [0.396]
1.197 [0.549]
Number of Countries Number of Observations
15 211
7 108
9 103
Note: The number of countries does not sum up to 15 for the entire sample as Sweden changes its legislation during the time period of observation
Table 6a: Regression Results and Robustness for pmp measures Dependent variable: Natural logarithm of cadaveric donors per million population (1)
(2)
(3)
(4)
(5)
(6)
-0.063 (0.167)
0.129 (0.157)
-0.286** (0.148)
-0.261** (0.115)
-0.259** (0.123)
-0.366** (0.173)
0.316** (0.084)
0.398** (0.096)
0.286** (0.093)
0.645** (0.101)
0.305** (0.097)
Catholic Country
0.349** (0.141)
0.475** (0.063)
0.418** (0.081)
0.348** (0.133)
Log of MVA and CVD deaths (per 10,000 pop.)
0.264 (0.183)
Presumed Consent Log GDP per Capita
Log of Blood Donations (per 1,000 pop)
R-square
0.023
0.304
34
0.438
0.371* (0.205) 0.316 (0.343)
-0.287 (0.419)
0.491 (0.406)
0.453
0.623
0.509
Table 6b: Regression Results and Robustness for 20-75 deaths measures Dependent variable: Natural logarithm of cadaveric donors per relevant potential deaths between 20 and 75 years old (1)
(2)
(3)
(4)
(5)
(6)
-0.207 (0.224)
0.058 (0.152)
-0.286** (0.084)
-0.371** (0.151)
-0.259** (0.123)
-0.364** (0.184)
0.533** (0.086)
0.579** (0.068)
0.645** (0.078)
0.645** (0.101)
0.668** (0.103)
Catholic Country
0.463** (0.067)
0.398** (0.153)
0.418** (0.081)
0.366** (0.149)
Log of MVA and CVD deaths (per 10,000 pop.)
0.264 (0.183)
0.238 (0.189)
Presumed Consent
Log GDP per Capita
Log of Blood Donations (per 1,000 pop)
R-square
0.064
0.564
0.603
0.579
0.234 (0.219) -0.287 (0.419)
-0.119 (0.461)
0.623
0.634
Table 6c: Regression Results and Robustness for MVTA deaths measures Dependent variable: Natural logarithm of cadaveric donors per relevant potential MVA deaths
Presumed Consent
(1)
(2)
(3)
(4)
(5)
(6)
-0.567** (0.199)
-0.302 (0.188)
-0.351 (0.271)
-0.444* (0.259)
-0.362** (0.166)
-0.443 (0.261)
0.463** (0.106)
0.469** (0.105)
0.333** (0.095)
0.326** (0.106)
0.333** (0.095)
0.064 (0.283)
0.116 (0.026)
Log GDP per Capita Catholic Country Log of Blood Donations (per 1,000 pop)
R-square
0.318
0.589
0.591
0.116 (0.262)
0.642* (0.374)
0.607 (0.397)
0.642 (0.385)
0.673
0.648
0.672
Table 7: IV Regression Results
Dependent variable: Natural logarithm of cadaveric donors per million population (3) (1) (2) Legislation: Presumed consent Wealth & health expenditures: Log GDP per capita
Religious beliefs: Catholic country
-0.452** (0.221)
-0.256 (0.261)
-0.491 (0.314)
0.372** (0.106)
0.338** (0.126)
0.334** (0.085)
0.515** (0.251)
0.682** (0.266)
Potential donors: Log of MVA & CVD deaths (per 1000 pop.)
0.232 (0.197)
0.483** (0.193)
R-squared Number of observations
0.324 193
0.298 149
35
0.312 145
Table 8: Refusal Rates by Families (Source: ONT, Etablissement Francais des Greffes)
Year
France
1996 1997 1998 1999 2000 2001 2002
30% 32% 32% 31% 32% 35% 32%
Refusal by fam ily PC Spain Italy
IC Australia
24.8% 21.7% 21.3% 21.8% 22.0% 23.4% 20.0%
59%-86% 58%-85% 57%-86% 59%-81% 56%-84% 57%-85% 56%-82%
24% 21% 18% 19% 23% 21% 20%
Note: UK Transplant started to measure the refusal rate of the families in 2004 (42% refusal rate)
Table 9: Amount Raised for French Telethon (source: AFM) Year 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Figures of French Telethon Amount received in million dollars Per capita Amount 29.47 0.54 28.32 0.51 40.68 0.73 46.23 0.83 36.62 0.65 47.58 0.84 55.26 0.97 57.17 1.00 56.41 0.98 56.53 0.98 62.65 1.08 70.02 1.20 72.15 1.23 74.39 1.26 83.40 1.41 85.60 1.44 91.60 1.53 104.70 1.74
Table 10: Knowledge about the Telecommunication Market Has a Ground Line
Total Does not know anything about the market Knows some names Knows some companies but does not really know the different offers Knows the sector rather well but sometimes lost Knows the sector well
Uses an Alternative Provider for Long Distance Calls
No Ground Line
Cell Phone
No Cell Phone
Internet
No Internet
17%
17%
8%
11%
9%
30%
5%
20%
46%
47%
48%
40%
44%
50%
33%
50%
20%
19%
20%
26%
24%
12%
31%
17%
13%
12%
17%
19%
17%
7%
21%
11%
4%
4%
6%
3%
6%
1%
10%
3%
36
