Testing the impact of phone calls on service delivery in India

|

This page in:

Before last summer’s planting season, the Indian state of Telangana proposed a new way to help farmers – give them money.  Many of Telangana’s farmers are subsistence farmers who often struggle to find the money they need for seeds, fertilizer, and pesticides. Through a new program called Rythu Bandhu, or “Friend of the Farmer,”  the state government promised to help farmers cover their costs by giving them 4,000 Rupees (about US$55) per acre before the summer and the winter planting seasons.


 

Figuring out who would get the money was easy – the government had registered all landholders the previous year. The challenge was ensuring that the government, which expected to hand out about US$1.8 billion annually through this program, could get checks to everyone before the onset of the monsoon season. This meant delivering physical checks to the 5.7 million farmers in the state within one month.

Enter a team of impact evaluation researchers funded by the Strategic Impact Evaluation Fund as part of its portfolio of low-cost, rapid  studies known as nimble evaluations. The team planned a simple randomized control trial that collected information on how the program was working and tested a way to improve it. As part of the testing, the government created a call center that called farmers in a randomly selected set of jurisdictions to track if they had received their money or had any problems, such as being asked for bribes. The civil servants tasked with handing out the checks in those areas were told that their performance would be monitored through these phone calls. By comparing two groups of farmers – those whose checks were delivered by civil servants who knew about the phone monitoring to those not being monitored-- the research team could test the impact of monitoring local bureaucrats via phone and whether telling local civil servants about the calls would spur them to do a better job getting the checks to beneficiaries.

“We were having discussions with the Telangana government about different program evaluations, and the key issue was how to make sure the cash would reach people successfully,” said Burak Eskici, a member of the research team and now a project director at University of California, San Diego.

J-PAL, the global impact evaluation research center, knew that cash transfers could be hard to do successfully. A few years earlier, they tracked a program in a different part of India that replaced wheat handouts to poor families with money deposited in their bank accounts. On paper, almost 100 percent of the transfers were made. But when the researchers followed up, some 30 percent of beneficiaries reported that they hadn’t received the money, either because they didn’t know about the program or because the money may have gone to the wrong bank account. The government didn’t want the same thing to happen here.
 
In Telangana, both the program and the evaluation moved fast. Check deliveries started in early May 2018, and by May 23rd, farmers had received and cashed 2.8 million checks. The call center made calls to some 50,000 beneficiaries over two weeks in late May and early June. By early September, the research team provided Telangana officials with evidence they could use when they launched a second payment round before the October planting season.

And how did the program work out? Overall, when civil servants knew they were being monitored, farmers were more likely to get their checks: 84.3 percent of farmers in the monitored group received their checks, compared with 83 percent of farmers in the unmonitored group. The 1.3 percentage point difference may seem small, but in the context of the total amount of money being handed out, it translated into an additional 17,000 poor farmers receiving transfers and US$1.0 million reaching farmers thanks to the phone call monitoring. The cash transfers also went out more quickly, with US$3.9 million more reaching farmers before the start of monsoon season.
 
“Unlike many evaluations that work well at a small scale but cannot be implemented more widely, this evaluation shows that phone-based monitoring can succeed even at the scale of an entire state,” said Karthik Muralidharan, an economist at University of California, San Diego, and member of the research team.
 
The government was pleased with the experience and satisfied that it got value for money. Paying the call center cost about US$36,000, a fraction of the money that made it to the right farmers thanks to the monitoring.
Overall, the cost-per-dollar of benefits delivered to beneficiaries was 3.6 cents, which is lower than the administrative cost of almost any anti-poverty program for which such data is available.
 
Now, officials in Telangana want to know if they can use this method to monitor other programs, possibly starting with a new initiative to bring piped water to more communities. The research team, meanwhile, sees possibilities to expand both use of phone monitoring to other states, something they are currently working on with officials.
 
SIEF has a portfolio of 20 nimble evaluations designed to improve policymaking by providing governments and stakeholders and with evidence to improve programs  when it’s needed. For more information, visit our evaluation page.

Authors

Aliza Marcus

Senior Communications Officer

Join the Conversation