Title:
|
Encouraging payment for services : experimental and
quasi-experimental evidence from developing countries
|
This thesis presents three studies which examine the decisions of individuals to pay for clean water
and public services. The three impact evaluations, conducted in India and Guatemala, demonstrate that
small changes to the way services are provided and presented to individuals can have a large impact on
demand for them.
The first chapter looks at whether tariffs discourage the use of NGO funded water quality projects
in Telengana, India. The estimation strategy takes advantage of panel data from water purification plants
where the tariff for 20 litres of water changed. In this setting, a one standard deviation decrease in price
is estimated to cause a six percent increase in uptake. The second paper evaluates an adaptation to
similar NGO water projects which increased the hours clean drinking water was accessible for, from 4
hours a day to 24. The results from the cluster randomised controlled trial show that the intervention
increased demand for clean water by 22 percent.
The final chapter presents findings from a nationwide randomised controlled trial in Guatemala that
tested various letter reminders to promote tax compliance. The results show that whilst all letters
increased the rate of declaration, only two of the letters, which were adapted using behavioural design,
were successful at increasing the rate of payment and the average amount paid per letter received. These
two letters more than tripled tax receipts compared to the no letter, or the original letter. The best
performing letter, if sent to all taxpayers in the sample, would have generated an estimated US$757,837
of extra tax revenue in 11 weeks compared to no letter, over 35 times the cost of sending the letters.
Overall, the thesis provides evidence that both economic and behavioural dimensions affect
payment for clean water and tax, and so both must be considered when development practitioners and
policy makers design 'customer facing' programmes. The methodology used in the three studies
highlights that it is possible to rigorously evaluate such programmatic design decisions at low cost using
experimental and quasi-experimental methods.
|