Using lab-in-the-field experiments to predict and understand new product take-up: evidence from helping Filipino migrants send remittances for education

Many policy interventions combine several features that we think may all potentially be key for the results we are trying to achieve. For example, conditional cash transfers typically combine giving cash to the household, some message about the importance of health and education, some condition that requires the household to go to health clinics or kids to attend schools, and details such as who receives the cash (mother or father), how they receive it (directly paid to bank accounts or paid in cash), and the frequency of receipt. One approach to understanding the importance of these different features has been to randomize these features, offering a multitude of different policies (e.g. Berk’s work in Malawi, or the recent Give Directly trials).

Such an approach has obvious advantages, but requires very large samples to give sufficient power to compare the effects of different features, and often requires randomizing at the level of a cluster to avoid confusing recipients with multiple versions of the same policy being implemented in the same small area. It is therefore interesting to also think of alternative approaches which may give some insights into which key features of a product or policy matter most, without having to actually introduce competing versions of the product.

In a new working paper, I worked with Giuseppe de Arcangelis, Majlinda Joxhe, Erwin Tiongson and Dean Yang to use a lab-in-the-field approach to unbundle which features matter most in a new product designed to allow migrants to direct some of their remittances for education purposes. A new Finance & PSD Impact note discusses some of the policy implications of this proof of concept trial – here I concentrate on the use of the lab-in-the-field experiment to unbundle the features of the product.

The new product: EduPay
                Using a Development Innovation Ventures (DIV) grant from USAID, we worked with the Bank of the Philippine Islands (BPI) and the Philippine Association of Private Schools, Colleges, and Universities (PAPSCU) to develop and pilot a new remittance product, called EduPay. This was piloted among Filipino migrants in Rome, Italy.
                EduPay allows migrants to directly send money to schools in the Philippines to pay for school fees, and receive back attendance reports and grades from the school. The idea behind this is to overcome information asymmetries and differences in preferences between the migrant and relatives in the Philippines: the migrant may be reluctant to send money to pay for a relative’s schooling if they are unsure about whether it will really be used for this purpose.
                Now this product contains several features that potentially may drive product usage and ultimately development effects. It provides a form of soft commitment device, by labeling these remittances as being for education purposes. It makes this commitment harder and lowers transaction costs by directing these payments direct to the school. And it provides information back on the outcome of this remittance through attendance and grade reports. In theory all may be important, but it is unclear which features matter most.

The lab-in-the-field experiment
We carried out a modified dictator game to understand which features matter most. A sample of 501 Filipino migrants in Rome with relatives aged 5-22 in the Philippines were asked to allocate 1000 euros between themselves and as many people in the Philippines as they liked. They were told a lottery would be held, with the choice of the winner being implemented exactly as specified, so they should make this allocation choice seriously. Each migrant made this choice under four different scenarios, with the order of these scenarios randomized:
1)            Standard remitting: a simple dictator game, in which they choose how much to send to people in the Philippines.
2)            Education labeling: the option to label funding as being for the education of someone in the household.
3)            Direct payment: the option to have money sent straight to a school to pay directly for school expenses.
4)            Direct payment + performance monitoring: the option to send money to the school and also receive attendance and grades directly back.
We also randomly allocated individuals into three groups, differing in what the remaining family would be told about the choices the migrant faced, in order to test whether choices differed as we varied the degree of information asymmetry.
The Figure below shows the outcome of these choices in terms of the total amount of money they choose to remit (whether for education purposes or not):


Migrants remit about 15% more when given the option for education labeling, but then only an insignificant additional 2% when the option to directly pay the school is available, and no additional amount more when it is also possible to monitor the child’s school performance. This suggests the key feature is the labeling rather than direct payment.

We then offered the EduPay product to the participants in this lab-in-the-field experiment to see whether their choices in the game affecting their willingness to use the new product. We find that:
 

• Individuals who choose to remit more money via the direct payment option during the game are significantly more likely to take up the EduPay product. A one standard deviation increase in the amount chosen to be remitted via direct payment option in the game is associated with an 8.7 percentage point higher take-up rate, which represents a 32 percent increase relative to the mean take-up rate of 27.1 percent.
• We then use within-person differences in the amount they choose to remit under the different options to see whether this explains take-up. We find that a one standard deviation increase in the differential amount remitted when given the option for education labeling (versus standard remitting) is associated with a 5.1 percentage point increase in the likelihood of take-up, representing a 19 percent increase relative to the mean take-up rate. However, individuals who remit more when given the direct payment option than when given just the education labeling option are no more likely to use the product.

These lab-in-the-field results suggest that the demand for this new product is driven largely by its ability to label remittances as being for an education use, rather than from the hard commitment or monitoring features. This is useful from both a theoretical perspective, for understanding how the product works; as well as from a logistical perspective – we experienced a number of difficulties in trying to pay the schools directly as they were often very slow to respond with payment details, so if it turns out that this component is not needed, it suggests the product can be much easier to deploy.

These results therefore suggest that lab-in-the-field experiments have the potential to be useful ways of understanding which features of a new policy or product may matter more in situations where it is not feasible to randomly offer multiple versions of the policy.