DN DEBATT 10/11. This year's prize in economic science in memory of Alfred Nobel has been widely acclaimed in the media. Commentators have gone so far as to call this research a ”godsend” to poverty reduction and that it will help ”restore the relevance of the profession”. Others have argued that it is now possible to determine with certainty what works and does not work when it comes to aid to developing countries. The question is whether this enthusiasm is justified? writes Bo Rothstein.
En utskrift från Dagens Nyheter, 2019-12-07 03:59
Artikelns ursprungsadress: https://www.dn.se/debatt/the-nobel-prize-in-economics-ethics-and-corruption/
This article was published in the op-ed section in main Swedish morning daily newspaper Dagens Nyheter on November 10, 2019.
Researchers Esther Duflo, Abhijit Banerjee and Michael Kremer have received the award mainly for their application of a research method which is called ”randomized control trials” which has for long been standard in medical research. This method has quickly become widespread within development economics and is considered by its representatives to be the absolute best way to ensure causation. In its motivation for the prize, the Royal Swedish Academy of Sciences provides some examples.
• The main idea is that by drawing lots, certain public organizations that provide services, such as schools in a developing country, are selected and in these the employment conditions (”the treatment”) for the teachers are changed to see if this improves teaching compared to the schools were this ”treatment” is not provided.
• Another example is when you by lottery selects a group of schools where the students are give medication against parasitic diseases and then compare their school results and health conditions with the students who went to the schools where they are not given this medication.
• A third example is when you select a group of schools by lottery where you install cameras that check if teachers come to their classes and compare them with schools where you have not installed such camera surveillance.
• A fourth example is when, through the lottery's help, one selects a number of water wells that one protects against pollution and then compares the health status of these users with the users who have to use unprotected water.
There has been a lot of discussion among economists about the value of this research, not least about the problem of external validity, that is, if the policy recommendations that follows form the experiments will work when applied generally.
However, there are three difficult problems with this research that have not got the attention I think they deserve. As is well known from medical research, all forms of experimentation with humans involve ethical problems. One might wonder about the ethics when these researchers are going to explain what happened to the children who became seriously ill as a result of their lack of luck with the draw in these kinds of experiments.
1 Since the prize is given in Sweden, formally by the Swedish Nobel Foundation but in practice by the Royal Swedish Academy of Sciences, it seems justified to check what the Swedish laws about research ethics state. A main rule of Swedish legislation regarding ethical review of research is that the researchers have an obligation to inform the test subjects that participation in the experiment is voluntary and that they have the right at any time to cancel their participation. These rules have not been followed in many of the kinds of experiments that have now been awarded this very prestigious prize. Participation in many of these experiments cannot be seen as voluntary since children in poor countries often cannot change school. One can easily imagine that in these settings, it can be very difficult to get to a different water source than the one closest to where you live.
Moreover, in the case of children under the age of fifteen, the Swedish law stipulates that researchers must obtain the custodian's permission, but adds that even if one obtains the consent of the custodians, ”research must not be carried out if a research person under the age of 15 realizes what it means to his or her part and opposes it being performed”. I have not been able to find any statement in the research articles referred to in the Academy’s motivation of the prize that indicates that the researchers in question have informed and obtained the children's consent to be experimented with. For the most part, the term ”consent” cannot be found in these articles.
2 To this is added a second ethical problem, namely that researchers under Swedish law must also inform the subjects about the purpose of the research they are asked to take part in. This is also difficult to do in this type of randomized control trail research because of a well-known problem within the social sciences named the ”Hawthorne effect”. The name comes from a factory in the United States where, in the 1930s, the effect of improving employees' work environment was investigated. It turned out that it was the very experience of being observed by researchers that improved the productivity of employees, not at all the efforts made for improving their working conditions. For the research discussed here, the requirement to inform about the purpose of the research usually implies that the value of the study perishes because those who are part of the experiment and are told the purpose, as a result of this knowledge, will change their behavior.
For example, teachers who know that they are part of an experiment to investigate what will increase their ambitions in teaching will of course increase their efforts during the experiment itself, regardless of whether they are, for example, camera-monitored or not. Parents who know that their children are participating in an experiment investigating the effect of giving medicine for serious illnesses will, if they have been unlucky in the draw, try to get hold of the medication given to the children who have been more fortunate in draw. The Hawthorne problem seems to be the reason why many researchers in this field choose not to inform the subjects. According to a recently published study, almost half of the scientific articles that uses this method that are published in the leading journals in the field lack information on whether the subjects were informed. In addition, two of the award winners (Banherjee and Duflo) highlight in an article the benefits of not informing those that are to be experimented with.
It should be stated that it is possible to apply to ethical review authorities for exemptions from informing the purpose. Practice and laws are different in different countries but something tells me that the rules about research ethics are not as stringent in many developing countries as they are in Sweden and probably in many other western countries. If this assumption is correct, this research actually has a strong flavor of neo-colonialism.
3 The problem with the ”Hawthorne effect” leads to a third problem, namely that the method in question is not suitable for investigating what in my opinion (and the opinion of many other researchers), is considered the main problem behind poverty in developing countries, namely corruption. In, for example, healthcare, the police and the schools, the requirements for extensive bribery are numerous in many developing countries. From a theoretical point of view, it would be possible to investigate what are effective measures against bribery by randomly selecting certain health care clinics, schools and police stations and trying out different measures that might work. But if you, following the requirement inform the staff that you are going to investigate which measures work against bribery, you will of course get massive ”Hawthorne” effects. Because they know about the purpose of the experiment, the staff will stop requesting bribes regardless of the measures being tested and thus the results from the experiment in question will of no value.
In addition, if the researcher would be so lucky to come up with result based on this method that they can show works against corruption, the policy recommendation that would follow will have to be implemented by state authorities in these countries that are often severely corrupt. The effect is then very likely to be zero. This is thus research where the ambition to get ”perfect measures” trumps the ambition to get results that can work in the real world. Thus, these researchers have as much chance of finding a cure for poverty in developing countries as the drunken man looking for his lost keys where the street lamp's light reaches and not where he lost them.
It is not to the honor of the Royal Swedish Academy of Sciences (or the Swedish Nobel Foundation) to have awarded this prestigious prize to a type of research that could possibly be carried out in developing countries but which in all likelihood, for ethical reasons, would not be allow in their own country. In addition, not one word in the 40-page price motivation touches on the ethical issues in this experimentation with poor people in countries plagued by systemic corruption.