Rwanda: where even poverty data must toe Kagame’s line
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
In landlocked Rwanda, a small mountain state in the centre of Africa, strongman President Paul Kagame has overseen a miracle. From the ashes of a genocide in which some 800,000 people were killed in just 100 days, Mr Kagame has built a thriving economy, clamped down on corruption and lifted thousands of people out of poverty, official data show.
Since he became president in 2000, growth in gross domestic product has exceeded 7 per cent a year, infant mortality has halved and access to education and healthcare has soared. In turn, the country has become a darling of the development community.
The World Bank has committed more than $4bn to the country since the 1994 genocide and championed huge structural reforms in sectors including health, education and agriculture.
Even as some exiled former allies have questioned the government’s economic performance and criticism of Mr Kagame’s authoritarian tactics has mounted, the World Bank’s support has continued. In the past decade, opposition parties have been squeezed out of the political system and dozens of regime opponents have been detained or died in suspicious circumstances.
Mr Kagame was elected for a third term in 2017 with 99 per cent of the vote. In 2018, World Bank funding to Kigali more than doubled to a record $545m.
Yet a Financial Times analysis of government statistics has found that the data look to have been misrepresented on at least one occasion, casting doubt on both the strength of the proclaimed economic miracle and the integrity of Rwanda’s relationship with its biggest donor.
The Rwandan government says poverty has reduced progressively since 2001 in the country of 12m people. But according to an FT analysis of survey data published by the Rwandan bureau of statistics, poverty increased during at least one important period — the run-up to a referendum in 2015 that allowed Mr Kagame to extend his then 15-year rule for up to another two decades.
Opposition politicians say the country’s poverty level is part of a much bigger deception over economic progress in which donors, keen to laud Rwanda as a success story, have become complicit.
“[The government is] trying to convey that we are developing so they can hide what is really going on,” says Diane Rwigara, who was jailed for 12 months after she challenged Mr Kagame for the presidency in 2017. “When you come here as a visitor all is put in place to impress you, but the reality is well hidden. You have to live it to believe it.”
A small number of academics first challenged Rwanda’s poverty statistics in 2015, leading the country to revise its analysis in 2016 and the World Bank to publish its own response last year. The academics’ findings, some of which have been published by the Review of African Political Economy, are compelling, independent experts say, but have been drowned out by the strength of Rwanda and the World Bank’s comprehensive denials.
However, the FT analysis of the survey’s more than 14,000 data points and interviews with academics shows that rising prices for Rwandan families meant poverty most likely increased between 2010 and 2014.
Rwanda and the World Bank reject the FT’s findings, insisting their own calculations are accurate.
“Rwanda’s performance in poverty reduction . . . is unequivocally real,” Yusuf Murangwa, director-general of the National Institute of Statistics of Rwanda, says. Progress was further corroborated, he adds, by the positive trend in other areas including financial inclusion, the expansion of tax receipts and the results of the country’s demographic and health surveys. “None of that would have been possible if poverty had actually been rising, as alleged,” he says.
Critics such as David Himbara, who was head of Mr Kagame’s strategy and policy unit until he fled the country in 2010 over what he says was his reluctance to massage official data, insist that poverty levels are just one of many manipulated figures in a regime where even statistics must toe the party line.
“Every number for Kagame matters whether it is politics or economics, and that is the way he convinces the donors to look away from his repression and rather concentrate on economic development,” says Mr Himbara, speaking from Canada, where he lives in self-imposed exile. “[The donors] can look away from the authoritarian side because he promises to deliver on the economy, so he has to keep showing that the numbers are great.”
The NISR assesses poverty levels via a regular household survey that asks what a representative sample of Rwandans are consuming on an annual basis. The NISR then compares the costs of those goods to a national poverty line.
The result of its fourth such survey was published in August 2015 as Mr Kagame was preparing for the referendum. Every assessment since 2001 had recorded a drop in the number of poor in the country and the government expected progress to be made again.
The 2015 report did not disappoint. The proportion of the country living in poverty had fallen from 44.9 per cent of the population in 2011 to 39.1 per cent in 2014, it said.
But the finding was immediately challenged. Measuring levels of poverty, like calculating other economic indicators, is a complex process in which different assumptions need to be made about factors ranging from the type and value of the products consumed by the population to how prices are likely to have changed over time and geography.
On this occasion, Rwanda’s statisticians had decided to update the basket of goods used to calculate the poverty line for the first time since 2001. The adjustment was necessary, the NISR said, because of changes in how people were living and eating.
While countries often update the way their poverty line is calculated, NISR did not initially make the same adjustments to the results of the previous survey, rendering the government’s comparison of poverty levels between 2011 and 2014 flawed, experts such as the Belgian academic Filip Reyntjens said.
In response, Rwanda analysed the results of the survey again in 2016, publishing a new report in which it used a cost of living index to adjust the results of the 2011 study to 2014 prices. The result, Rwanda said, was the same — the poverty rate had fallen.
“The key conclusion — that poverty fell substantially between 2010/11 and 2013/14 — is robust: it holds true,” Mr Murangwa said in the report. In fact, poverty had fallen by 6.9 percentage points, even further than first thought, he added.
The FT analysis of the same data contradicts that finding, suggesting there has been a consistent attempt since 2015 to misrepresent the results.
According to the FT calculations it is only possible to show a poverty decline of 6.9 percentage points if the mean value of the NISR’s cost of living index was 4.7 per cent. Put more simply, poverty could only have fallen by such a large margin if average prices for the poorest 40 per cent of households increased by 4.7 per cent or less between January 2011 and January 2014.
Four academics contacted by the FT say they believed prices in rural communities increased by significantly more than 4.7 per cent during the period in question and that it looked like a low estimation of price rises had been used to skew the results and imply a reduction in poverty that did not exist.
Sam Desiere, a senior researcher at Belgium’s University of Leuven who has studied Rwanda’s poverty statistics, says average prices probably increased by at least 30 per cent between 2011 and 2014 based on his analysis of price data included in Rwanda’s household survey. That conclusion would imply that poverty increased by about 6.6 percentage points, according to the FT analysis. “The higher the inflation rate,” says Mr Desiere, “the more poverty increases.”
Rwanda denies that it has misrepresented the results. “The claim that poverty in Rwanda increased between 2011 and 2014 is wrong,” says Mr Murangwa.
The NISR later added that it was “inappropriate” to use a mean of the cost of living index to understand how it adjusted prices between the two surveys but declined to provide any further details about the rates it used.
The subjectivity of the assumptions economists must make to estimate and compare poverty levels makes it very difficult to reach a definitive conclusion. But even Rwanda’s own consultants disputed the government’s view that poverty had declined.
UK-based consultancy Oxford Policy Management was hired by the NISR to help it complete the poverty analysis in 2015, as it had done on each of the three previous surveys.
OPM’s team found that poverty had increased by at least 6 per cent but the finding was rejected by the Rwanda government, according to a person familiar with the company’s work for the NISR, who asked not to be named.
With the constitutional referendum approaching that December, officials were under pressure to show continued progress and there was no way an undesirable increase in poverty could be tolerated, the person says. After weeks of debate over Rwanda’s proposed new methodology, OPM ended the contract and did not sign off the final report, the person adds.
OPM confirmed it was contracted by NISR in 2015 to help analyse the results of the survey but declined to comment further due to client confidentiality.
Inside the World Bank, the poverty report had also split opinion. While the country office officially backed the findings, five World Bank staff sounded an alarm in a letter to the organisation’s senior leadership, details of which are published here for the first time.
“Recent developments point to potentially serious reputational risks for the Bank if its Rwanda operations continue in the current trajectory,” the officials wrote in the letter dated 1 December 1 2015 and addressed to Jim Yong Kim, then World Bank president, Makhtar Diop, then vice-president for Africa, and Diarietou Gaye, then country director for Eritrea, Kenya and Rwanda.
The group, calling itself “Professionals for Truth in Aid”, signalled their concern over what they called “the manipulation of official statistics and failure to provide reliable data openly”.
The group received no response and decided to go no further. “We sent it anonymously because we were so afraid of retaliation [from the Rwandan authorities],” says one of its authors, who asked not to be identified.
The World Bank says it cannot verify the authenticity of the letter or confirm whether it was received by Mr Kim, Mr Diop or Ms Gaye because it was not sent using the institution’s official whistleblowing systems.
Regarding OPM’s alleged concern over the reliability of the poverty data the World Bank said it was “aware of the discussions [between OPM and NISR in 2015] but was not privy to the deliberations between the government and its consultant”.
The World Bank has twice defended the government’s poverty analysis: in a 25-page report in September 2018 called Revisiting the Poverty Trend in Rwanda and in a public statement in April.
In both papers, the World Bank said the wide discrepancy between the consumer price index — which puts inflation for 2011-14 at 23 per cent — and the much lower index used by the NISR required further investigation but backed the conclusion that poverty had dropped. “We have not found any clear sign of errors or manipulations,” it said.
One of the problems in reaching a definitive answer on what the survey data show is the variety of options for counting inflation. Rwanda, for example, argues that the CPI inflation rate of 23 per cent is skewed towards urban prices, and that the cost of foodstuffs in rural areas — 80 per cent of Rwanda’s population lives in rural communities — was increasing more slowly from 2011 to 2014.
The government’s overhaul of its agriculture sector from 2008 led to increased productivity, which improved the standard of living for rural families, it says.
But An Ansoms, a professor in development studies at the University of Louvain in Belgium, says the policy had the opposite impact. Under the programme, which was supported by the World Bank, farmers were instructed to grow specific crops in specific regions.
The programme led to increased productivity, but also “huge price inflation” for those crops that people did not grow, Ms Ansoms says. “In perfect economic theory, you would have perfectly integrated markets and then you would get economies of scale effects . . . but that is not what happened.”
The rising prices led to increased scarcity for many smallholder farmers in the rural areas where Ms Ansoms conducted field work between 2010 and 2014, she says. “Even if people got a quite decent price for their maize,” she says, “the bargaining power of the salary that they got was not at all at the same level of what they produced before.”
Letters in response to this article:
No single ‘best way’ to estimate poverty works at all times in every context / From Thomas O’Brien, World Bank Africa Region
Rwanda’s progress is real and open to scrutiny / From Uzziel Ndagijimana, Minister of Finance and Economic Planning, Rwanda