Using geospatial data to measure local government quality in Africa
Geospatial data doesn't have to be used for mapping environmental and economic factors. It can also be used to examine (and address) local government corruption.
Geospatial data—and any data for that matter—is a means to an end.
In other words, data is something that helps us unlock analyses, and to understand things in a better way.
So what I find exciting is how we can use geospatial data to provide insights into things that really matter.
My whole journey into geospatial data came about from spending some time in Uganda in 2016.
I was there as part of a World Bank project trying to understand the following graph—i.e. why Uganda’s manufacturing industry couldn’t seem to grow.
Manufacturing was thought to be key to promoting economic development and better living standards.
This is because manufacturing:
Increases productivity: Equipping labour with machines massively increases the productivity of many unskilled or semi-skilled workers.
Addresses skills shortfalls: Manufacturing enables countries with a shortage of skilled labour to leverage machines to increase output.
Creates tradeable goods: Manufactured goods can be easily traded across borders compared to services. E.g. it’s easier to export furniture than it is to export haircuts. This makes the market for manufactured goods much larger (i.e. it can be global, whereas services may only have a local market).
And when we increase production and output, we increase GDP.
Doubling down on manufacturing is also what the ‘Asian Tigers’ like South Korea and Singapore had managed to do so successfully during the 1990s:
So, what were the factors that were stifling the manufacturing industry in Uganda? And how did geospatial data help?
The barriers to manufacturing growth: hypotheses
Going into Uganda, we had a bunch of hypotheses about why manufacturing growth had stagnated.
These included:
High exchange rates — the cost of imported raw materials was growing due to a depreciation in Uganda’s exchange rate. This made it more expensive for manufacturing firms to build stuff.
High interest rates — the cost of borrowing was enormous, which made it difficult for manufacturing firms to take on debt to buy machinery, etc.
Electricity shortages — the electricity supply in Uganda hadn’t been stable, and electricity surges can fry machines and other equipment.
Land tenure issues — problems with the land tenure system meant that multiple people may have the same claim to a plot of land. This could be a barrier to the expansion of factories.
We set about interviewing over a hundred firms of all sizes to try and validate these hypotheses.
Of course, many of the issues above were mentioned as contributing factors. However, one particular issue emerged that we hadn’t expected: the quality of local governments.
Many firms we spoke with had complained about inefficiencies and even corruption of local governments. This was a pretty interesting finding, particularly because Uganda, like many countries in Africa, had devolved several responsibilities from the national government to local governments. The World Bank was also increasingly working with local governments. So it was important to examine this issue in more detail.
I turned to the literature to try and understand this issue of local government quality within Africa in more detail. I was surprised to find how few quantitative studies had been written about local government quality, than national government quality within Africa. This was despite a profusion of such studies in Europe.
So this ultimately led to me undertaking a PhD to examine this issue in more detail.
Mapping local government quality with geospatial data
The first challenge in understanding local government quality was how to quantify it.
Luckily, Afrobarometer, a survey that covered 200,000 respondents over 37 African countries, includes a bunch of questions about local government quality. Some of these questions include:
How much do you trust your local government?
How many local government officials do you think are involved in corruption?
Do you approve/disapprove of your local government official’s performance over the last 12 months?
These responses are geocoded at the household level. So this makes it possible for us to use geospatial techniques to aggregate the data, and map local government quality in a country.
Using these survey responses, my co-authors (Neil Lee and Andrés Rodriguez-Pose) and I created an index of local government quality:
This was essentially an index that combined various factors that make up the ‘quality’ of local governments:
the level of corruption in local governments
the level of trust people had in their local government
the quality of services delivered by local governments
However, the issue with metrics like this is that they’re inherently subjective. In other words, we built the index using people’s perceptions of their local governments.
I would have preferred more ‘objective’ measures of government quality (e.g. audited budget papers from local governments). However, the availability and quality of official or ‘objective’ datasets rapidly deteriorate once we examine things at the local level.
Assessing the impact of local government quality on economic development
Mapping local government quality was the first step.
But the major question is “so what”? How does this actually impact sub-national regions?
In order to answer this question, we wanted to study the impact of local government quality on sub-national GDP.
Despite its flaws, GDP is a good metric that’s highly correlated with:
living standards
educational attainment,
the quality of healthcare, and so on.
So if we wanted one metric to assess general well-being, then GDP is a good bet.
The problem, which I’ve discussed previously, is that official sub-national GDP data doesn’t exist for many countries in the world. As a result, we used nightlights as a proxy for sub-national GDP:
Ultimately, we find that local government quality directly impacts a sub-national region’s GDP. In other words, using geospatial data, we were able to find that issues in local governments ultimately impact the living standard of people living in those areas.
By mapping local government quality and demonstrating how it can impact residents, we now have a data-driven method that enables national governments or organizations like the World Bank to target and improve local governments with lower ratings.
This could be through assigning them more funding, implementing training programs, etc.
How local government quality is being addressed in Uganda
That said, some of the most impressive results I’ve seen in terms of improving local government quality have been done at the grassroots, community level.
A Ugandan non-profit, SEMA, has been doing some great work in the country.
SEMA focusses on collecting feedback from citizens who use services provided by local governments. This could be a person visiting an office to obtain a new ID, or when visiting a court or health centre.
SEMA does this by sending volunteers to conduct on-site interviews with citizens. These data collectors are located at the exits of local government offices. Upon leaving the office, citizens are asked about their experiences, wait time, and overall satisfaction.
Information gathered through the surveys and devices is aggregated into a monthly one-page report, which is delivered to the head of each local government office.
The report provides a grade, shows performance compared to the previous month, compares the office to other offices, and explains where an office performed well and where it needs to improve.
SEMA’s hypothesis is that if citizen feedback is presented to local government offices regularly and in an easy-to-understand format, this will increase accountability, and incentivise service delivery improvements while simultaneously providing a way to monitor and evaluate the quality of services over time. By comparing local offices to each other, this also creates a competition effect.
The results of this intervention are pretty interesting:
Creating incentives: By making government performance transparent and fostering a competitive environment among offices, SEMA’s program has created new incentives for improvement, leading a shift towards better service delivery.
Cultural and systemic change: SEMA’s work is creating a broader cultural change towards better service delivery. This is because most of the offices SEMA collects data from have increased their performance over time.
However, despite these promising results, there are a couple of other factors to take into consideration:
Challenges with structural corruption: Tackling deep-rooted corruption through citizen feedback alone has been promising, but has its limitations. Corruption can often occur at higher levels that aren’t directly visible to citizens.
Resource constraints: SEMA’s work has highlighted structural issues like under-resourcing and low pay, which are significant barriers to improving service delivery. Incentives alone can’t address these types of constraints.
Conclusion
Through this example, we’re able to see how:
Geospatial data can be used to map important issues such as local government corruption at a local level,
Geospatial data can be used to assess the impact of these issues for example by examining how local government corruption can impact standards of living and GDP, and
Targeted interventions can be used to address these issues at a local level, such as SEMA’s grassroots efforts to improve local government quality and service delivery.
Sometimes we get so drawn into the weeds about the latest dataset or data processing methodology, that we can forget about how we can apply these tools to examine important issues around the world.