I warmly recommend “Poor Economics,” a book by M.I.T. economics professors Abhijit Banerjee and Esther Duflo, published by PublicAffairs, New York. The subtitle of the book sounds provocative: “A radical rethinking of the way to fight global poverty.”
This book presents the results of a long list of studies based on a research method that, even if it is not new in development economics, is not yet commonly used and is certainly not familiar outside the scientific community. This research method uses Randomized Controlled Trials (RCTs). They are used to evaluate the effects of policies.
Why do researchers need a special method to evaluate the effects of policies? When a policy is implemented, it is often implemented on the whole population. In that case, it is not possible to gauge the effect of the policy since one does not have any group to compare the population to. Sometimes, the policy is not implemented on the whole population but only some part of the population (young people, children, unemployed, inhabitants of city C). It is possible to evaluate how the treated population is doing compared to the rest of the population. However, how can we be sure that the differences we are observing between the two groups come from the policy and not from other factors impacting this part of the population (imagine you compare city C to city D, but in city C a rather unusual event occurs that impacted its inhabitants at the same time as the policy).
RCTs deal with those two problems. They consist of implementing large scale experiments, during which a large sample of people is randomly divided in two (or more) subgroups, and only one of the two subgroups is “treated”, that is, is faced with the evaluated policy. The simple principle is that if we observe a difference in some relevant characteristic between the two groups after the experiment, then this effect qualifies as an effect of the policy. This is due to the fact that the two populations are randomly assigned to one of the two groups.
For a nice introduction on RCTs check out Esther Duflo’s TED talk below:
The authors themselves have conducted a lot of RCTs. They report the results of their experiments, as well as those of many other RCTs. The outcome is that many new ideas emerge, and many received ideas happen to be deeply challenged. In this post, I would like to give two examples. The first one deals with a study of conditional cash transfers. The second one deals with microfinance institutions.
Conditional cash transfers
Mexico was the first country to implement conditional cash transfers dedicated to giving incentives to parents to send their children to schools. Poor parents were given some amounts of money, conditionally on their accepting to send their children to school and to seek preventive health care. The effects have been immediate: school attendance significantly increased among the children of those families. As a consequence, conditional cash transfer programs have been implemented in many other Latin American countries and elsewhere in Africa and Asia.
The paradox of this story is that the Mexican experiment cannot be taken as a proof that conditional cash transfers are effective in convincing parents to send their children to schools. This is what RCTs in Malawi and Morocco, among others, have shown. In those studies, samples of poor families were divided in three groups. Families in group 1 were not “treated.” Families in group 2 received some unconditional cash transfers. Families in group 3 received some cash transfers conditional on sending their children to schools. The first finding is that school attendance remained low in group 1. The second and most surprising finding is that there was no difference in the increase in school attendance between groups 2 and 3. That proves that cash transfers alone do give incentive to parents to send their children to schools, and not the fact that they are conditional.
There is a lot of evidence that poor families are cash constrained, that is, that they would like to borrow some money to spend on consumption or investment goods and would be able to reimburse it but do not find anybody to lend them money at a reasonable interest rate. Facing this evidence, microfinance institutions have been started in Asia and then widespread everywhere else. Those institutions typically lend money to groups of people, who are then held collectively responsible to reimburse the loans. This collective responsibility lowers the risk of default, and, therefore, decreases the cost of lending, so that interest rates can be drastically lower.
Microfinance has rapidly been viewed as a key component of any program of poverty alleviation. The problem, however, was that only anecdotal evidence was available to document their effects, so that it was difficult to claim that the observed effects were to be attributed to the microfinance institutions. On the other hand, anecdotal evidence of negative side effects of microfinance (like farmers unable to repay loans committing suicide) was also available.
The authors carried out a RCT study in Hyderabad, the capital city of Andhra Pradesh, South-East India. They took advantage of the development of a microfinance institution in the city. Out of the 104 new neighborhoods which the institution wanted to reach, 52 were randomly chosen for the institution to enter, whereas the 52 others were left unaffected. After more than a year, the researchers came back to compare the changes that occurred in the two samples of neighborhoods.
The evidence is that the access to cheap loans did affect and improve the lives of the poor families in the first set of neighborhoods, but did not radically transform it. For instance, it is true that more than 7 percent of families in the first set of neighborhoods succeeded in starting a new business thanks to the availability of cheap loans, but no less than 5 percent of families in the second set of neighborhoods succeeded in starting a new business, too, without access to any microfinance institution.
This book is full of fascinating stories of failures and successes in the fight against poverty. Clearly, the original analyses of the reasons why one policy fails and the other one succeeds force us to rethink the design of policies.