High-Frequency Data Collection: A New Breed of Household Surveys

A Precise, Timely View of Poverty

Measuring poverty in the developing world is no easy task, as the perennial debate over the World Bank’s release of poverty statistics can attest.

One challenge is the lag in household data. The most recent and comprehensive global poverty statistics end in 2010, already three years ago.

There is a reason for such delays: Many countries lack the resources or capacity to conduct and publish results of surveys – the collection of data the World Bank depends on to compile poverty statistics.

This, in turn, means that governments and their partners may not be able to target responses to food or economic crises where the help is most needed.

Enter high-frequency surveys, a new and emerging breed of income and consumption surveys that use economic modeling or wireless technology to capture the poverty picture in real time. 

“High-frequency data is doing to economics what genetics did to medicine,” said Marcelo Giugale, the World Bank’s director of economic policy and policy reduction programs in Africa. “We always had economics before which allowed us to do policy and try to solve people’s problems, but with high-frequency data we’re able to do things we couldn't do before.”

Mobile Phone Surveys Give Instant Answers

In 2009, as the global financial crisis took hold, phones started ringing at the World Bank in Washington. Government officials were calling to ask how the economic downturn was affecting the poor in their countries, and how they could best respond.

Amparo Ballivian, a lead economist and formerly with the Latin American and Caribbean region, was asked by her director to summon her team and come up with some answers.

“We had to tell him the truth: We didn’t have any data,” Ballivian recalled.

A discussion began in her group about how to get data more frequently, Ballivian said, so they would “be able to answer similar requests in the future when we have similar crises.”

The idea was hatched to distribute mobile phones that surveyors could call or text on a regular basis to ask questions, 10 during each interview, about employment, violence, hunger, illness and access to social services. Cell phone coverage in Latin America is almost ubiquitous, which made the region a promising experimental ground.

After initial on-the-ground tests in Nicaragua and Peru in 2010, Ballivian’s team contracted with the Gallup organization to conduct a six-month pilot study in Peru and Honduras that began in October 2011 and January 2012, respectively. Brightstar, a distributor for the wireless phone industry, donated 1,000 phones that were given to families that needed a phone to participate.

Gallup ran monthly surveys of 1,500 households and the project yielded some interesting findings.

While incentives such as free airtime appeared to boost public interest in the mobile phone surveys, for example, the level of incentive (more versus less free airtime) did not. Participation can also vary significantly from one country to the next.

Results from the survey were analyzed during summer 2012. The plan was to scale up the mobile survey project to additional countries in Latin America and the Caribbean later that year.

Mobile phone surveys will never replace comprehensive household surveys, but they can offer valuable information about what's happening at that moment in people's lives. How many meals did they have today? Has anybody in the family been sick. Were the kids able to attend school in the last week?

There is a definite need for new and inexpensive technology and methodology - especially in countries with significant poor populations, said Jaime Saavedra, director of the World Bank's Poverty Reduction & Equity group. While a typical Latin American country conducted eight annual household surveys in the past decade, he noted, African countries managed two.

An alternative to mobile phones: New methods for estimating poverty

As the push grows to get more timely information about human conditions, World Bank teams are also developing a new methodology to help them estimate consumption levels for years when no regular household surveys are conducted. The model, which uses complex formulas to predict data, is adjusted each time a new household survey is conducted to make sure the margin of error remains acceptable.

This high-frequency, or “proxy,” survey methodology was tested in Bangladesh in 2011. The country was chosen for the project because it conducts regular household surveys only every five years, and because the World Bank economist spearheading the methodology, Nobuo Yoshida, already had a good working relationship with officials in Bangladesh.

Yoshida was comfortable with the results from the Bangladesh pilot project, and shared them with government officials in Dhaka in mid-2012. If the meeting goes well, the the government of Bangladesh will incorporate the new methodology in its National Strategy of Developing Statistics, making it part of its regular country monitoring process.

“Our method may not be perfect, but there are other methods out there that also have shortcomings, including the fact that they are more expensive and time-consuming,” Yoshida said. “We’re using existing data so in that sense, this is very cost-effective.”

Yoshida’s hope is that the proxy survey methodology, as well as surveys using mobile phones and other innovative technology, will benefit nations as they seek to measure poverty in a timely manner. The ultimate goal for poverty economists such as Giugale, Ballivian and Yoshida is to help governments improve human conditions for their citizens in a world where nearly 1.3 billion people still live in desperate poverty.