Africans are right to be sceptical of those who pay little heed to contextual, on-the-ground research, writes Morten Jerven, author of Africa: Why Economists Get It Wrong.
For the past two decades, mainstream economists who study African economic growth have been trying to explain something that never happened. Economists have focused almost exclusively on one question: why has economic growth failed in Africa? Answering this erroneous question is not only an academic pastime; the perspective has made its way into popular domains. Most famously, in 2000, the front page of The Economist depicted Africa as a hopeless continent that was unable to experience economic growth and development. In a special report on the continent, the magazine asked “Does Africa have some inherent character flaw that keeps it backward and incapable of development?”
Only ten years later, The Economist had an entirely different front-page report about Africa as a hopeful continent that was on the rise. How could they have got it so wrong the first time? In 2000, Johannesburg’s weekly business magazine Financial Mail pointed out that, in 1997, just three years earlier, The Economist had written that “sub-Saharan Africa is in better shape than it has been for a generation”. The Financial Mail asked: “Do the editors of The Economist have a character flaw that makes them incapable of consistent judgement?”
As Bob Dylan told us, you do not need a weatherman to tell you which way the wind blows. One should perhaps not be too surprised that The Economist’s judgement varies and turns with the current. In the year 2000, the editors were looking back at two decades of Western news coverage about Africa that focused on famines, civil wars, and failure. The magazine’s perspective in 2000 was very much a child of its own time. However, it was a rash judgement that ignored certain fundamentals in African economic development. From the “weathermen” in the story of Africa’s development – actual economists – one would expect a judgement that relies on long-term patterns and history and stands the test of time. Yet, economists continue to get Africa wrong.
When the editors at The Economist declared that Africa was hopeless, they were not just putting their finger in the air to see which way the wind was blowing, they were taking inspiration from the consensus in the economic literature at the time. Since the 1990s, almost all economic studies of Africa had focused on explaining why there has been a “chronic failure of growth” on that continent.
In 2007, University of Oxford economist Paul Collier identified the “bottom billion”, the population of the world that (according to Collier) live in countries that do not experience economic growth. He identified just under 60 countries that he calls Africa+.
In retrospect, what is so striking is that this statement was made just after a period of rapid economic growth since the mid-1990s in the very economies Collier was talking about. The majority of the economies that Collier described as chronic failures had been growing for more than a decade. Many countries grew before, after and even during the time when Collier was writing his book. So my question is: how could economists miss decades of economic growth?
Madness in the method
The answer is in their methods. We need to go back to a seminal paper by Robert J. Barro in 1991. He proposed a model that would later be the workhorse in what has been called the “regression industry”. He wanted to explore the determinants of economic growth using a global sample of countries. On one side of the equation, he put average growth of countries, and then on the right hand side of the equation he proposed a set of variables that he suggested could explain the variation in average country growth rates. Yet, after having added a range of such variables to capture differences in policy regimes, education levels and political unrest, Barro still had an unexplained residual. He also entered an African continent “dummy variable”. A dummy variable takes the value 1 if the country was African and 0 if the country was not African. He found a significant negative African dummy variable, and his interpretation was that the analysis had not yet fully captured the characteristics of a “typical” African country.
What ensued was a decade of research into eliminating the “African dummy”. The objective was to find a variable that captured this character flaw in African countries. Many variables were suggested and found to correlate with slow average growth – a survey found that no less than 145 variables had been suggested as determinants of growth. The problem remains that correlation does not imply causation. Let us take one variable that was suggested: high aid dependence. So one could plausibly argue that high aid dependence is bad for growth because it means that states are externally oriented, that they do not focus on delivering public goods in return for taxes collected and so forth. That’s all fine, but does it stand up as a cause of slow growth on the African continent?
It does not. It breaks the first rule. Cause should precede effect. If you look at the data you will see that recorded rates of official development assistance as a share of GDP are only relatively high since the 1980s and 1990s, while before that aid as a share of GDP was relatively low. Thus, it does not make sense that aid caused slow growth – rather the opposite, it appears that recorded high aid dependence is best explained as an outcome of slow growth in the 1980s and 1990s. Other correlates such as budget deficits, high black market premiums or subjective measures of the quality of the political institutions suffered from exactly the same error. Observations originating in the 1980s and 1990s were taken to explain the whole period.
The fundamental error in these models is that they focused solely on explaining the average shortfall of growth in Africa. This meant that the recorded decline in the 1980s and stagnation in the 1990s completely overshadowed the gains made in the 1960s and 1970s. Not only did this create an erroneous and overly pessimistic picture of economic performance in African economies, it also conveniently circumvented some of the difficult questions for the orthodox economic literature. How can we explain that so-called “bad” economic policies in the 1960s and 1970s coincided with good economic performance, whereas the introduction of “good” economic policies and political governance in the 1980s and 1990s is correlated with economic stagnation and political turmoil? This economic analysis gave support to the liberal policy package enforced in the 1980s and 1990s, but the record shows that growth only returned when world economic conditions improved in the late 1990s.
Economists failed to examine these contradictory patterns, which would have entailed questioning some of the basic assumptions in the models and the validity of the evidence. Rather, the mainstream economic literature went ahead and accepted “chronic failure of growth” as a stylised fact. In the second generation of the growth literature, the central research question was no longer, whether “bad” policy was correlated with poor economic performance. The question was rather, what kind of character flaw these countries had that caused them to persistently follow policies that were bad for growth.
This literature inspired The Economist editorial in 2000. Economists used a range of different datasets to pick up some of Africa’s distinctness – was it high ethnic fragmentation, colonial legacies, the slave trade or incidence of malaria or other diseases. Currently, the mainstream economic literature is able to explain why African economies are not growing, why they are failing and why they are stuck in poverty traps. The problem (for economists’ theories but not for African countries) is, African economies have been growing for about two decades. There is nothing new about this growth. It is particularly frustrating, and it surely stands in the way of objective evaluation, that the narratives in African economic development switch from one extreme to the other so swiftly. The truth lies somewhere between the “miracles” and “tragedies”. It is nothing short of stunning that in a matter of a few years the most famous phrase relating to African economies for economists has turned from “Bottom Billion” to “Africa Rising”.
Some economists and political scientists systematically do research on Africa that results in misleading findings. Good comparisons should be reciprocal – that means that you can take both sides of the comparison as the norm. Instead, the economic growth literature has used the subtraction approach – so that we are able to explain lack of growth in Africa with respect to “lack of governance” or “lack of social capital” or any other variation of a characteristic where African countries are found to be different. As a result, we have a literature that can explain how African economies have not performed, instead of a literature explaining what actually happened.
Another related disturbing trend is the increasing distance between the “expert” observer and the observed African economy. While country studies used to be the norm in the 1970s and 1980s, this has given way to cross-country growth regressions with global datasets. With downloaded data, there are few ways of double-checking whether reality and the data sets correspond. Moreover, much of the “governance” evidence is based on using data sets that are explicitly surveying subjective opinions held by outsiders. As a result, much of the data that economists use to make judgements about Africa is plain wrong, flawed or shallow.
In essence, most of the macro literature using an African sample studies “economics” not “economies”. The primary interest has been coming up with law-like general statements. That often means that relevance and local applicability is sacrificed. To be told that “history and institutions matter” and then subsequently be presented with what one commentator called “Wikipedia with regressions” is indeed a paradox. Ironically, if the literature is correct it means that global sample regressions should be abandoned in favour of deep contextual studies of history and institutions. The challenge is to keep economics relevant for the economies it purports to study.