Recipes for economic growth, ingredients and all

Everyone knows that Africa’s economies have recently taken off. Rising Africa has become a cliché phrase in economic articles. But what does the rise exactly look like and what contributes to that? 

With the help of World Bank income data, it is easy to see that the Sub-Saharan Africa has indeed experienced over two-fold increase in incomes during the last decade. Below, I have marked with a blue line the Gross National Income (GNI) levels per capita during the first decade of 2000. We see that the income levels have increased significantly. In order to put income levels of last decade into context, I have added a red line marking the average GNI levels during the three previous decades, from 1962 to 1999, and a green line marking the previous peak income in 1979. See below: we can see that Sub-Saharan Africa has never seen so high GNI levels per capita before.

While it is true that income levels alone are not sufficient to illustrate the economic situation of the entire continent consisting of 54 countries, we can nevertheless conclude few things from that chart below: first, despite the complaints of Africanists and African scholars against The Economist’s simplistic labelling, the simple comparison of Sub-Saharan income levels indeed seems to justify tagging Africa “hopeless” in 2000, “rising” in 2011.

Now, it is much more difficult to say what exactly contributes to the growth. In an attempt to narrow my focus in Africa further down, I have chosen only one country – Rwanda – for my economic analysis, for which I have used 1984-2009 data from World Bank. 

But data alone are not enough. One also needs some models regarding how to use that data. If you wish, economic growth models can be considered like recipes for economic growth, each with particular ingredients. Regression analysis, which can be considered as cooking (again, to bring a lousy parallel) allows me to make a causal claim about how these ingredients in those recipes condition the outcome. Or – what recipe with what ingredients best cooks the economic growth?

All three models, which I picked: namely the Rostow’ stages of growth, the Harrod-Domar model, and neo-classical model – attempt to give the best formula for the economic development. As you see below, they have slightly different ingredients, and different number of ingredients, as well.

Which one of them offers the best recipe for growth? 

Rostow’ stages of growth	The Harrod-Domar model	Neo-Classical Model
1.    Savings, 2.    Investment, 3.    Population growth;	1.  Level of savings, 2.  Productivity of capital (or national capital–output ratio, in inverse relationship);	1.  Capital accumulation (in other sources savings rate), 2.  Population growth (or labour growth), 3.  Productivity, and 4.  Technological progress.

One problem was to find correct, matching data for all those ingredients. Although World Bank offers a wide range of data – starting from the classic GDP to the number of mobile cellular subscriptions and to the number of newborns protected against tetanus to the rate of condom use among youth, I nevertheless had difficulty finding exactly matching (and consistent) data. Like trying to cook exotic cuisine at home, I had to replace some ingredients.

For example, while preparing an old-fashioned Rostow’ stages of growth, which required savings, investment and population growth, instead I used ‘gross domestic savings’, and ‘foreign direct investment’, both measured in current US dollars. I also used size of the total population for the third ingredient: population growth.

As a result of running scatterplots and regressions analyses to explain how do these three ingredients contribute to the GDP growth, I found that savings, population and investment explain impressingly large share of the growth outcome: 91.8 per cent!

This means that countries seeking to boost their GDP should try to find ways to increase their level of savings and investment in addition letting their population grow.

I found this outcome quite surprising as the Rostow’s model is considered to be out-dated. Old-fashioned, as I said. Moreover, it must be mentioned that the single ingredients in this model did not quite correlate perfectly to the GDP. See, for example, how the graph above illustrates correlation of population size to GDP: it is not linear. There are periods, where GDP goes down while the population grows. This tells us that African 'demographic dividend' alone is not sufficient to guarantee economic growth.

Second model on my list, the Harrod–Domar model has only two ingredients. Will it beat the Rostow's model in explaining GDP growth?  Harrod-Domar model tries to explain economy's growth rate in terms of the level of savings and productivity of capital. Therefore, it only has two components. Again, I found it is hard to measure productivity. That is why I replaced ‘productivity of capital’ with ‘national capital / output ratio’, which was available in World Bank dataset. The peculiar thing about measuring capital / output ration is that - the its smaller value, the higher the productivity is: its value correlates negatively to the growth. 

For example, compare the 100-dollar-output produced by 1000 people (which is 10), with the same output worth of 100 dollars produced by fewer, 900 people. The outcome is smaller, 9. That can be shown on the graph below – the higher the productivity (the smaller the value), the bigger the GDP.

Compared to the previous recipe, the Harrod-Domar model explains much smaller share of GDP, as I found after running regressions with the two ingredients. My smart computer program SPSS calculated that only 73 per cent of economic growth is explained by those two ingredients. This tells us that one needs more ingredients (and likely another model) to cook a good economic growth.

And the last one...

Now, lets look at the last, neo-classical model as it is called. For this "recipe" there are different ingredients listed in different sources. It consists of productivity, population growth (or labour growth), capital accumulation (in other sources savings rate), and technological progress.  For consistency, I was using total labour force / GDP ratio as a proxy for productivity as I did with the previous model, and the same indicator for population growth as in Rostow’s model. But instead of required capital accumulation, I found a ‘gross capital formation’ indicator in the World Bank databank.

But how to measure the required ‘technological progress’? There is no exact indicator provided in the Worldbank data for Rwanda. I had to decide: which one is better replacement: the high-tech export, measured in current US dollars, or number of scientific and technical journal articles published annually in Rwanda?

In order to decide, I tested them in correlation to the GDP… How do they relate to eachother?

Since the slightly better match to GDP growth was offered by the number of scientific articles published in Rwanda (see below),  I choose this indicator for the fourth required ingredient of technological progress. (It is quite funny, if you think that the number of journal articles can have anything to do with economic growth.) My computer program did not laugh though – it liked all the ingredients in this recipe, and as a result of regressions, I can claim that together they explain an astonishing 98 per cent of economic growth (while the result is statistically significant)!

This means that based on Rwandan data from three decades, it can be suggested that the leaders who look for recipes for economic growth, could consider the guidance offered by neoclassical model. 

Therefore, capital accumulation (or rate of savings), population (or labour force) growth, productivity, and technological progress will most likely lead to the best outcome in economic growth.

Comparing the models and the outcomes, I realised that the recipe with more ingredients will explain the outcome in GDP growth the best. Neo-classical model has four variables and it provided the highest percentage of GDP growth explained. Therefore, we can conlcude that there is no single magic bullet for prosperity – instead, improvement in a wide variety of factors is necessary.

But is it possible to claim which of those ingredients or independent variables is most powerful?  Yes, based on regression analysis, the rise in productivity seems to be contributing to the GDP growth the best – every unit reduced in labour force /GDP ratio seems to be contributing the biggest rise in GDP growth of all other independent variables. Therefore, it seems clearly: the higher the productivity, the higher the GDP growth. 

-xxx-