A Pakistani official set out to prevent bureaucrats from demanding bribes for providing basic public services. The solution: citizen feedback through mobile phones.
- By Mohammad Omar MasudMohammad Omar Masud is pursuing a PhD in urban policy at the Massachusetts Institute of Technology.
In early 2008, an irate government official walked into the office of Zubair Bhatti, the district coordination officer (DCO) of Jhang district, in Pakistan’s Punjab province. The official complained that, when he tried to register his property, the land registry employee had asked him for a bribe. Bhatti immediately conducted a surprise inspection of the district registry and launched a formal corruption investigation into the behavior of the accused employee. But he worried that few of Jhang’s 3.3 million residents were lucky enough to have the DCO intercede on their behalf when facing such demands from bureaucrats.
Punjab’s 36 district administrations handled services such as property registrations and residency certificates, as well as health, education, and income assistance programs. Delivery of these services and programs involved hundreds of transactions with citizens every day. DCOs were responsible for the delivery of public services in the districts but did not have the time and resources to spot-check every transaction to determine whether public employees were soliciting bribes.
Like other districts in the province, Jhang had a system that enabled the DCO to initiate formal inquiries on the basis of citizen complaints. But the inquiry process was cumbersome and required frequent personal appearances by the complainant before the inquiry officer. Ordinary citizens did not have the motivation and time to go through the process, and many feared alienating local public officials. Public officials themselves had grown to distrust the public complaint system because of its susceptibility to spurious or malicious reporting. Official anti-corruption agencies were more interested in pursuing high-profile cases of corruption than in sifting through cases of petty wrongdoing encountered by ordinary citizens.
Bhatti decided to experiment by ordering officials of the property registration office to note the cell phone numbers of citizens who used the service. He started calling random numbers from the list and asking the respondents about their experiences during their visits to the registry. He discovered that many visitors didn’t know the amount of the fees they were required to pay for registering property. Many applicants who lacked knowledge of the property registration process or were poor readers relied on the services of deed writers, who processed paperwork in exchange for payments that were often arbitrary and excessive. The informal system created the opportunity for financial collusion between the deed writers and public officials who registered the property.
Bhatti decided to expand the phone call experiment to other district services in Jhang. Government workers recorded the cell phone numbers of citizens visiting service facilities on paper forms and forwarded them at the end of the day to the DCO. Bhatti said he spent about 15 minutes a day calling citizens at random, jotting down any negative feedback that required action on his part. Making personal calls to citizens sent a powerful message that he and other high-level officials in the district could be trusted and wanted to help. He realized he could build credibility for the government by reaching out to citizens rather than waiting for disgruntled citizens to lodge formal complaints through an impersonal system.
The Jhang initiative drew national press coverage and caught the attention of Shahbaz Sharif, the chief minister of Punjab, who saw the program as a step toward addressing the petty corruption that infected service provision. According to Transparency International’s Global Corruption Barometer in 2010, half of all respondents reported paying bribes for delivery of public services during the previous year. Sharif wanted to expand the citizen feedback model to all of Punjab’s 36 districts to enable the government to identify specific areas of corruption and poor service delivery throughout the province.
When Bhatti left the civil service in 2009 for a job in the private sector, his departure resulted in the loss of an early champion of the phone-call experiment and stalled work on its further expansion. Still, public interest in the initiative remained. Pakistan’s leading newspaper, The News, published a series of articles that questioned why the government did not move forward with the “Jhang Model.” In September 2009, The Economist picked the story of Jhang for a report on mobile telecommunication in developing countries. International recognition and continued national media coverage kept the Jhang model in the public eye.
Sharif, the chief minister, viewed the project as a priority initiative, and he made it clear that he wanted the project scaled up quickly to reach thousands of citizens every day. The chief minister’s office even secured Bhatti’s services on a temporary basis from his employer to help manage the expansion.
Bhatti started by presenting his experiment in Jhang to other DCOs, many of whom he knew personally. Media coverage helped cultivate interest in the initiative among some DCOs. Six DCOs from Gujranwala and Bahawalpur divisions volunteered for the pilot phase. Rather than promoting the program as an electronic channel for citizen complaints, proponents described it as a model for proactive governance and highlighted the value of seeking unadulterated citizen feedback in identifying patterns of corruption and poor service delivery. Citizen feedback could enable DCOs to take immediate administrative action against public servants suspected of corruption and maladministration rather than going through the laborious task of trying to build criminal cases, sometimes on scanty evidence.
Developing the citizen feedback model was a process of trial and error. The six pilot districts started to collect the cell phone numbers of citizens who used selected district services. Initially, the districts sent lists of numbers to Bhatti, who made random calls and collected feedback for the district coordination officers, identifying patterns of reported corruption and invalid numbers. After hearing recordings of Bhatti’s calls, the DCOs soon started placing the calls themselves. Their experiences left a positive impression on both ends of the communication.
In Gujranwala, one of the pilot districts, when citizens reported problems in the property registration branch, the DCO initiated anti-corruption investigations that resulted in the arrest of some of the staff at the office. Media coverage of the incident helped spread word of the new program and explain to the public how the system worked. The volume of calls increased, and four more districts joined.
As the program scaled up, it became clear that the manual process of collecting information and citizen feedback was too slow and labor-intensive. In June 2010 the chief minister transferred the operation of what was then called the Citizen Feedback Monitoring Program, or CFMP, to the provincial implementation and coordination wing to streamline the process. The Punjab Information Technology Board, an autonomous body set up by the province’s government, provided valuable support by introducing Web-based entry of citizen cell phone numbers in district offices and developing a detailed Web-based form to capture citizen feedback.
As expected, the pilot program encountered some problems. Despite the detailed online feedback form, DCOs who placed manual calls often failed to record the citizen feedback online. In addition to their regular duties, DCOs were expected to call about 5 percent of the telephone numbers collected daily. They struggled with the growing work burden.
Bhatti, who at the time was working for the World Bank’s Islamabad office, found a way to help. Each year, the World Bank hosted a competition from its innovation fund, which supported pilot projects that increased transparency and access to information. Developed in partnership with the World Bank and inspired by the CFMP pilot the provincial government proposed a project titled “Proactive Governance: the Punjab Model.” The project was successful in securing a $100,000 grant from the innovation fund for technical assistance. The grant helped pay for work to fix some of the early technical design problems the CFMP pilot had revealed.
In October 2010, the chief minister’s secretariat brought CFMP back under its control to maintain momentum. Noor Mengal, an officer in the chief minister’s secretariat, took over logistics and pushed districts for prompt reporting on citizen cell phone numbers, calls made by DCOs, and overall progress.
Money from the World Bank’s innovation grant enabled the hiring of several young computer science graduates to help strengthen and expand the CFMP. Asim Fayaz, a computer science graduate from Lahore University of Management Sciences, was the lead consultant recruited through the innovation grant; he partnered with Fasieh Mehta from Punjab Information Technology Board. The two provided a valuable interface between the software developers and technology vendors who designed the system and the government workers who used it.
In January 2011, the technology board contracted with the Pakistan Telecommunication Company to operate a call center to handle the growing scale and scope of the CFMP. Almost immediately, the program ran into a problem. Although the company had submitted the lowest bid, it had no experience in providing call center services. The company had to recruit and train a staff, develop scripts for call agents, build an online system for entering information from the phone conversations, and create an online dashboard for capturing and displaying citizen feedback. The company’s Chinese partner firm, which was designing the online dashboard, found it difficult to understand the local context and requirements.
As progress slowed on the development of the call center, Mehta and Fayaz came up with an alternative: an automated system that could send text messages to citizens’ cell phones and request feedback. Citizens could reply by texting. The team developed a basic dashboard that could store, categorize, and display citizens’ cell phone numbers and their text replies.
The texting system had a significant cost advantage over the call-center system and provided comparable reach. A 160-character text message cost 80 paisa (less than 1 cent), while an average three-minute agent call cost 5 rupees (about 5 cents). Convenience and effectiveness also favored the texting system. Because text messages did not require an immediate response, unlike live calls, citizens had time to consider their answers and discuss the messages with others. In addition, text replies could be stored more easily and accurately for analysis than audio recordings of phone conversations.
By July 2011, 15 districts had joined the project. The text message feedback system had recorded 5,000 messages and was growing fast. The numbers indicated citizens were responding to the program.
The program had collected enough citizen feedback to justify a province-wide rollout to Punjab’s chief minister and the provincial cabinet. Unconventionally, the presentation to the cabinet highlighted the positive feedback received from citizens about CFMP rather than dwelling on corruption numbers that were always open to controversy. The cabinet was shown actual text messages from citizens in which they thanked the chief minister and said they were grateful for the respect they felt the government was showing to them by asking for their opinions. The presentation also showed how citizen feedback could improve governance by pinpointing corruption all the way down the service facility level. The Cabinet approved a full rollout.
In 2012 the Punjab Information Technology Board opted for a different partner company to run the call center and negotiated a contract with TRG, the country’s largest call-center firm. The new call center used so-called robocalls that were recorded in the chief minister’s voice to convey the impression that he took a personal interest in each citizen’s opinions. The robocalls, which could reach large numbers of people quickly, informed the citizens about incoming text messages soliciting their feedback. The call center divided its work between monitoring the integrity of the cell phone data provided by districts (such as filtering invalid or spurious numbers), collecting and classifying text message replies, and following up with agent calls where the replies indicated the possibility of corruption. The project team developed an easily recognizable four-digit code from which it sent the text messages, and the provincial government launched an information campaign around the four-digit code to inform citizens about the importance of CFMP and the value of their responses.
In 2012, Nabeel Awan, who had been one of the first DCOs to pilot the citizen feedback model, assumed responsibility for the coordination and implementation of the CFMP in the chief minister’s office. Nabeel took a strong interest in analyzing the province-wide data. “Once or twice a week, [the team of consultants] would come to my office with CFMP reports on all districts,” Nabeel said in an interview. “We would go over them and identify unusual trends or patterns.”
The weekly meetings produced a reporting template to log the number of cell numbers collected by service providers in the districts, the percentage of invalid or repeated numbers, and the content of the feedback. Scant or no data from a particular facility or service center suggested implementation shortcomings. Unusual spikes in repeat numbers or invalid numbers from a service facility or district likely indicated attempts to avoid detection. Nabeel would report such anomalies to the chief secretary, the top civil servant in the province, who would press the appropriate DCO to respond.
By the end of 2012, the program had sent nearly 1.5 million text messages to citizens across the province. The citizen feedback program not only had become a monitoring and accountability tool but also was earning goodwill for the chief minister. In March 2013, the chief minister’s secretariat turned the citizen feedback model into an independent development project with its own budget, and placed it again under the provincial department for implementation and coordination wing. The chief minister also directed the chief secretary to make the CFMP review a regular agenda item in his monthly meetings with DCOs.
During its implementation, the citizen feedback program hit a number of hurdles, including leadership transitions, local political opposition, and contracting problems. Patronage was a significant issue in Pakistan. The CFMP sometimes brought DCOs into conflict with local political interests in cases involving negative citizen feedback about government workers who had been appointed through patronage.
But citizens’ comments empowered both the DCOs and the chief minister’s office to withstand political pressures. The notion that information on citizens’ perception about behavior of local public service providers was being instantly conveyed to the chief minister sent a powerful message to local politicians. The implied validity of the citizen reporting discouraged challenges against the new system and the chief minister. DCO Gulzar Shah explained, “I would show them a three-page-long list of negative citizen feedback coming out of the CFMP about a public servant, telling them that the chief minister has the same information, and they would not pursue the issue anymore.”
The CFMP also faced resistance from the independent deed writers who processed paperwork for property registration for citizens. In the Gujranwala district, deed writers approached the provincial high court claiming the DCO had no legal authority to ask for citizens’ cell phone numbers. In the end the court sided with the DCO, and the decision stifled any further efforts to challenge the program on legal grounds.
In May 2013, the Pakistan Muslim League won national elections, and the provincial government had new priorities, such as an energy crisis and concerns about terrorism, that sidelined the citizen feedback program. Nabeel left for a post in the prime minister’s office, which deprived the CFMP of a high-placed champion in provincial government.
By this time, the Punjab Information Technology Board under Umar Saif had taken over the job of sustaining the project. Saif met with public servants in the field to discuss how to improve the technology to help them in their work. In December 2013, the technology board created a helpline that linked the DCOs directly to program managers for questions and advice. In April 2014, CFMP team members began sending weekly text alerts to the DCOs indicating which service facilities in the district were reporting the lowest levels of data entry or the highest levels of negative citizen feedback. The team also sent text messages to thank citizens for their input.
What had started in 2008 as one DCO’s initiative to directly connect with citizens spread to all of Punjab, serving a population of more than 100 million. In April 2014, the World Bank commissioned an evaluation to measure the performance and impact of the CFMP. The phone-based survey covered more than 20,000 citizens who had used CFMP services since 2011. Preliminary results showed that 55 percent of those surveyed said overall service delivery had improved, 71 percent said staff attitude had improved, and 63 percent said timeliness of service delivery had improved. Although only 30 percent of those surveyed said the program had reduced corruption, 76 percent of the respondents said they still believed it would help reduce the problem in the future.
Successful implementation in Punjab encouraged other provinces to consider implementing their own versions of the CFMP. “Because the model was not excessively politicized, other provinces feel comfortable in adopting it,” said Fayaz, the lead IT consultant. The province of Khyber Pakhtunkhwa adopted the model in 2013, and in October 2014 the federal government replicated the CFMP platform for federal services such as passport and national identity card registration.
The underlying principle of the Citizen Feedback Monitoring Program – that the government should actively seek out citizens to ask them about their experiences with service delivery – was revolutionary in Punjab. The model allowed for both positive and negative feedback and was a way to systematically measure the quality of service delivery. The daily data collection provided a near-real-time picture of civil service performance, enabling the provincial government to react quickly and effectively to adverse trends in remote districts.
In late 2014, Bhatti, who developed the citizen outreach model, reflected: “If one big benefit comes out of it, it could be the simple idea that officers learn that they can just pick up the phone and call people up and get feedback. [About] anything.”
Photo credit: AAMIR QURESHI/AFP/Getty Images