16 April 2014.
By Kunal Sen.
Why are there such significant and persistent differences in living standards across countries? This is one of the most important and challenging areas of development policy. For individual countries in the developing world, extreme fluctuations in growth over a relatively short period of time can cause staggering changes in living standards. Income gains and losses of as much as three times GDP capita have affected whole populations.
Our new research aims to understand the causes of such economic growth (ESID Working Paper 26). First, we need to understand what growth is (ESID Handbook on Economic Growth). We show that economic growth in developing countries is an episodic phenomenon – massive discrete changes in growth are common in developing countries, with most developing countries experiencing distinct growth episodes: growth accelerations and decelerations or collapses.
For example, India was known for its ‘Hindu’ rate of growth till the late 1980s. Economic growth accelerated in 1993, and then again in 2002, and the combined income gain from these two growth accelerations was 3.7 trillion Purchasing Power Parity (PPP) dollars.
In contrast, Brazil was a ‘miracle’ country from 1967 to 1980, growing at 5.2 per cent per annum. But growth decelerated sharply in 1980 to essentially zero and stayed low until 2002. We estimate the net present value of the lost output from this slowing of growth to be 7.3 trillion PPP dollars, a loss of $61,000 per person.
The growth deceleration in Iran that lasted from 1976 to 1988 cost each citizen $146,643, a loss 11 times initial GDP per capita and over 5 trillion dollars.
Particularly tragic are the growth decelerations in Africa, where income fell from a particularly low base. For example, the growth deceleration in Malawi that began in 1978 cost each person cumulatively almost $10,000.
Thirty growth acceleration episodes in the post-World War II period have seen income gains (in net present value terms) of at least three times the initial level of GDP per capita at the beginning of the episode. In the same period, 32 growth deceleration episodes have seen income losses (again, in net present value terms) which have exceeded three times the initial GDP per capita at the beginning of the episode.
To estimate the loss and gain in income in growth episodes, we first identify structural breaks in GDP per capita for 125 countries, for the period 1950-2010. Once we know the timings of accelerations and decelerations in growth episodes, we estimate the magnitude of growth in these accelerations and decelerations. We compute the total net present value of the difference between the actual trajectory of output during the episode and the counter-factual of what the trajectory of output would have been in the absence of the onset of the growth episode. To obtain the counter-factual trajectory of output for each growth episode, we use a regression for each country/episode to allow ‘predicted’ growth to depend on a country’s initial GDP per capita, the episode period specific world average growth (to capture ‘global business cycles’) and the previous period’s rate of growth.
An important finding of our study is that ‘developed economies’ are at much, much less at risk of large growth decelerations. If we examine the 89 growth decelerations with negative losses, only two of these (Finland in 1985 and Italy in 2001) are rich industrial countries. Conversely, of the 90 ‘large’ or ‘medium’ magnitude decelerations, there are only six OECD episodes, nearly all slow-downs to moderate growth rates with large magnitude due to long duration (e.g. Greece 1973-2010, Spain 1974-2010, Austria 1979-2010, Switzerland 1974-2010). The majority of large growth decelerations have occurred in Sub-Saharan Africa and Latin America, and many are well known declines, often associated with political turmoil, conflict, and/or outright civil war (e.g. Iran 1976, Afghanistan 1986-94, Zaire 1989-2000, Nicaragua 1987-1995, Sierra Leone 1990-99, Uganda 1969-80, Ghana 1974-83, Somalia 1978-2010) or transition from central planning (e.g. Romania, Bulgaria, Albania). Some were extended slides into poverty (Malawi 1978-2002, Cote d’Ivoire 1978-2010).
The income gains and losses that we document in our study are too large to be explained by policy reforms, which cannot provide such big effects, even with pre-existing large distortions in the policy environment. Neither can they be explained by changes in fundamental determinants of long-run per capita income, such as institutions, which by their nature are slow-moving and sticky. What explains such staggering gains and losses in income over relatively short periods? This is the key question that ESID research on economic growth is currently addressing.
Update: See also the media briefing published at the Royal Economic Society.