How mobile text reminders earned Madagascar a 32,800% ROI in collecting unpaid taxes

Benjamin Franklin famously once said that "nothing can be said to be certain, except death and taxes." In developing countries, however, tax revenues are anything but certain. Madagascar is a prime example, with tax collection as a share of GDP at just under 11 percent. This is low even compared with countries of similar levels of economic development, and well below what the government could reasonably collect to fund much-needed public services, such as education, health and infrastructure. 

Poor compliance by citizens who owe taxes remains a major reason for Madagascar’s low tax collection. Madagascar’s government has therefore made increasing tax revenue collection a high priority in its strategy for promoting sustainable economic growth and addressing poverty.

Reforming a tax system can take decades. But small measures, implemented with the help of technology, can help tax authorities improve compliance.  Our team at the World Bank jointly conducted a field experiment with the Madagascar’s Directorate General for Taxation, to test whether simple text message reminders via mobile phones could increase compliance among late-tax filers.

We took a group of 15,885 late-income-tax filers and randomly assigned some of them to receive a series of messages reminding them to file a tax declaration and emphasizing various reasons to pay taxes. Late tax filers were told that they could avoid a late penalty by meeting an extended deadline and were given the link to the tax filing website. 

The results of the experiment were significant. In the control group, only 7.2% of late filers filed a tax return by the extended deadline cited in the SMS messages. This increased to 9.8% in the treatment groups who received SMS reminders. This might not sound like much, but for every dollar spent sending text messages, the tax authority collected an additional 329 dollars in revenues, making the intervention highly cost-effective.

In fact, the return on this particular investment was 32,800 percent! Although this increase in revenue is relatively small in absolute terms—around $375,000—it could be automatically integrated into the tax system. It also suggests that messaging may hold a lot of promise for cost-effectively increasing tax receipts even in developing country contexts.

We can track this increase over time in the following plot. The dashed lines show the time when the two text reminders were sent to the experimental treatment group. A clear gap in taxpaying opens up right after the second reminder message was sent.

While text reminders might not reshape tax collection in developing countries, our results suggest that not sending them may be leaving easy money on the table.