Student “data detective” untangles India’s rural wage-payment system
By Rebecca Cohen
Ford School of Public Policy
Through a paid research assistantship with the International Policy Center, Caroline Bach applied methods from her coursework to a study of how information could improve India’s rural employment program.
As a first-year graduate student at the University of Michigan Ford School of Public Policy, Caroline Bach’s research assistantship resembled an “archeological dig.”
Working with faculty adviser Yusuf Neggers, her work contributed to a broader research project that examines if low-cost digital tools can provide useful information to improve India’s Mahatma Gandhi National Rural Employment Guarantee program. The program provides eligible rural households with up to 100 days of paid manual work each year. Although wage payments are supposed to be processed within 15 days, they often take longer.
“If you’re vulnerable, not knowing when you’ll be paid for your work really undermines the value of that social protection program,” Neggers said.
Before the researchers could determine if digital tools, such as automated calls and reports, helped speed payments, they had to know whether the information was reaching the officials who could act on it.
Bach’s job was to identify those officials and measure whether they engaged with the information. That meant reviewing years of survey and program files.
The deeper Bach went into the data, the clearer it became that the records did not align.
“I expected to have to do some cleaning and management, but I found the data needed to be organized at multiple levels,” Bach said.
Official titles did not always capture who actually operated the program on the ground. Elected leaders named in the files could work through representatives, officials sometimes shared responsibilities and understaffed bureaucrats could cover multiple jurisdictions. Information missing from one record might appear elsewhere under another person’s name or role.
“On paper, there is a specific program administrator role,” said Jenna Allard, associate director of research and policy for Inclusion Economics at Yale University, who works with Neggers on the project. “In reality, based on the data that we collected, who fulfills these tasks was much more complex.”
“I started by asking which unit of analysis would best support future research,” Bach said. “Because respondents could serve at multiple levels—and locations could have no respondent, one, or several–we created two datasets. One was organized by respondent to identify who handled wage-payment work, and another organized by location to show where that responsibility existed.”
Drawing on quantitative methods and program evaluation concepts from her Ford School coursework, Bach verified summary statistics, flagged inconsistencies and unusual records, and investigated apparent gaps in surveys of local elected leaders and government officials. She also analyzed officials’ engagement using measures such as automated call-answer rates, listening time and link clicks.
“I learned to think critically about data quality and to break down complex questions into manageable, documented steps,” Bach said.
Bach operated like a “data detective.” “Even when individual records didn’t make sense, she would really narrow in on what the next steps were,” Allard said.
Bach’s questions also prompted the researchers to revisit assumptions they had taken for granted. Her analysis helped the team define engagement in measurable terms and better understand who was responsible for moving wage payments forward locally.
But reducing payment delays does not necessarily require giving officials more data. They may already receive more information than they can easily access, interpret or use. Earlier studies in Madhya Pradesh and Jharkhand, India, found that giving government officials simpler, more actionable information reduced the time it took to process wages.
The related experiment in Bihar, India, builds on that work by asking which officials should receive information, what would be most useful to them and whether they engage with it. Bach’s analysis helped the team identify the responsible officials and measure how they responded to the information they received.
A path toward international development
As Bach continues with the research assistantship in the 2026-27 academic year, she will move from preparing the data to studying whether providing officials with targeted information made a difference. The team plans to assess whether the intervention affected payment speed, the amount of work provided in participating communities and the distribution of employment opportunities among different groups.
That next phase will connect Bach’s detailed data work to a broader policy question: Can better information and accountability help an employment program deliver on its promise to vulnerable households?
The project has deepened Bach’s understanding of rural governance while strengthening her technical and analytical skills.
“My experience with [the International Policy Research Assistantship] reaffirmed my commitment to pursuing a career in international development informed by careful analysis, cross-cultural awareness, and continuous learning,” she said.
The International Policy Center’s International Policy Research Assistantship pairs enrolled students with faculty conducting international policy research and provides students with paid research experience. Depending on the project, positions are open to undergraduate and graduate students. The program is made possible through the generosity of the Weiser family.