What’s the premise of your book, Poor Numbers?
The book is a first-ever study of how African economies have been measured throughout history. I look at the information used to measure GDP per capita, including the data sources, the available data, and the methods used. And the book explains that there is a big knowledge problem: we know much less than we’d like to think about African economic development, based on the published numbers.
How did you get interested in this question?
I was preparing to do my PhD in economic history, and I was interested in the relationship between policy change and economic performance—changes in economic growth, for instance, before and after the structural adjustment programs of the 1980s and 1990s. I started looking at this for Tanzania, among other countries, and I could not make sense of the data. Different sources quoted very different growth rates, and the data did not match up with what I knew happened on the ground during that time. So, I was increas-ingly concerned that the numbers were painting a different picture from reality. I wanted to know: How do these international data-bases come up with their numbers? And how is GDP constructed for African economies?
So how is GDP constructed in these countries?
In theory, GDP is an expression of all value-added activities, or rather the production and/or consumption of all goods and services in one territory in one given year. You can get this number in one of three ways: the income approach, the expenditure approach, or the production approach. The income approach means adding up all wages, profits, and rents in the economy. The expenditure approach is consumption, plus investment, plus government, plus or minus exports and imports. Finally, you can add up the production of all goods and services by sector. That means you start in the agricultural sector, then move to manufacturing, mining, construction, all kinds of services, government, etc. That’s the production approach.
In practice, however, there is not enough information on private consumption on an annual basis to use the expenditure approach. And you do not have wage and profit information on this many small- and medium-scale operators for the income approach. So Af-rican statistical offices use the production approach. They go down the industrial classifications and say, “OK, for agriculture, we have some information on exports, but less on food. For manufacturing, we have information on some big operators, but we lack information on the small operators.” And so forth.
For the service sectors, there is very little information—there used to be more when the states were more involved in licensing and transport and wholesaling and so forth. So for some parts of the economy they have actual data, and for other parts they have to guess at it.
It’s important to know that there are parts of the economy that are recorded and parts of the economy that are unrecorded. And there is this gray area in the middle where guesses are used to try to arrive at a reasonable total.
Time for a Recount?
Countries calculate GDP starting from a base year, but when the base year becomes out of date, GDP data lose accuracy. The IMF recommends updating the base year every five years. [click for larger image]
Why have we gone along for decades now relying upon seemingly faulty data?
It’s useful to think about this as a knowledge problem and a governance problem. The knowledge problem is that we know less than we thought we did. But then there is the governance imperative—you need the data to make decisions. And often the governance imperative—the need for data—trumps the knowledge problem.
There is a longer trend as well. In the 1960s and 1970s, we knew less about poverty and households, but we knew more about economic growth and industry, labor, and services. But in the 1980s and 1990s, national statistical offices were getting fewer funds and less access to data, and therefore increasingly producing worse [economic] numbers. States disengaged a little bit from the importance of statistics and gave less priority to the statistical offices. Therefore, the production of these statistics went more into the hands of international organizations. It’s only recently that countries like Nigeria and Ghana have turned around and said, hey, what are our GDP statistics based on, we’re probably much richer than we think. States in Africa again care about economic growth, and they have put the numbers back on the agenda.
Ghana has already recalculated its GDP, and Nigeria is expected to do so. Is it reasonable to think that some countries in Africa are actually richer than their current GDP suggests?
It is for quite a few of the economies. Ghana hadn’t had a reasonable picture of its economy taken since 1993. They made a new up-to-date picture of the economy in 2010, and they found out that they were almost twice as rich as they thought they were.
To understand how a GDP revision can cause such a radical change, it’s important to understand what a base year is. A base year is the year the statistical office chooses as a starting year for its GDP series. Usually a base year is picked when the statistical office has more information on the economy than is normally available, such as data from a household, agricultural, or industrial survey. The information from these surveys is added to other administrative data to form a new GDP estimate. This total is then weighted by sector.
Sectors that were important in the base year will continue to appear important despite structural changes that may occur, while sectors that were unimportant or nonexistent will barely have an impact on GDP. When the base year is out of date, the GDP series becomes unreliable. The IMF recommends changing the base year every fifth year. In the case of Ghana, their previous base year was 1993. Obviously, the structure of the economy had changed radically since then. It turned out that since 1993 almost half of the Ghanaian economy had gone missing from the official count.
There are many other economies with base years that are really out of date. This means quite a few economies are probably richer than we think.
But this does not mean they are necessarily growing faster than we think. The knowledge problem is doubly biased. We know less about those economies that are poorer because they have more trouble getting an accurate measure of their economy. And [within economies] there is less information about the poorer population. We know more about the modern sector, we know more about imports and exports, but we know less about the small-scale and medium-scale sectors, we know less about what we call the informal economy, which by definition is unrecorded.
So my main message is that, yes, some of the economies today are probably much richer than we think, but economic growth today is biased toward picking up modern economic growth, external economic growth, while we know very little about how that relates to actual living standards for the average citizen.
It’s been suggested—in response to your book—that this problem of economic statistics is not necessarily a significant constraint on development and perhaps isn’t a priority. Why is it important to address this?
I think it’s hugely important. Of course I understand that GDP or economic growth is not all [we should be] interested in. There are many reasons why we should think about human rights, infant mortality, literacy. But the central banks cannot set their exchange rates, or make decisions on interest rates, based on infant mortality. You need to have some kind of information about what the economy is doing in order to make these decisions.
Moreover, it’s important to think about the national statistical office—GDP is just an analytical angle to figure out what the statistical office is able to do. The GDP is an appropriate way of looking at how much the state knows about itself. These statistical offices are unable to give the central bank information about how much cement was produced last year. We do not have labor statistics, although presidents are being elected or not elected on the basis of promising unemployment reduction and so forth. I don’t think the national statistical office is only supposed to measure GDP—I think GDP is a good way of showing that a very important statistic is seriously in disarray.
We have been on a development path where we have gotten more information on social and health indicators, which is good news. The bad news is that other kinds of information have not been available. If it’s true that Africa is rising now, a lot of investment banks and multinational companies would like to have some data to make their investment plans and so forth. I think it’s time for people in these countries to reassert themselves and say, OK, what are the data we need for our own development? And that means we need a rethink about what kind of information we are producing about development in Africa.