In his recent Energy in Africa post Euan Mearns made this statement:
One hypothesis I want to examine is that electricity is fundamental to GDP and GDP growth. Without it, individuals cannot create wealth. I was therefore expecting to see that electricity consumption should be correlated with GDP and growth.
There’s no doubt that electricity is fundamental to GDP growth and that wealth in our modern society cannot be created without it, but a key question is; which comes first? Does the electricity create the wealth, or does the wealth create the electricity, or is the linkage between the two so close that it’s impossible to say? This will be the second question this post addresses.
The first will be the question of the correlation between GDP growth and electricity consumption that Euan was expecting to see but which was not visible in the small sample of African countries he reviewed. But again there’s little doubt that such a relationship exists, and a larger sample should reveal it. Accordingly this post expands the sample size to 168 countries in order to determine how strong the relationship is and how it varies from place to place.
The data used in this review are from the following sources: Population data (mostly 2014 and 2015 estimates) are from Wikipedia . Nominal per-capita GPD data (2013 estimates) are from the UN and electricity consumption data (2012 estimates) are from EIA. It should be noted that the numbers in these data sets are not always the same as those published by other sources, such as the World Bank, the IMF and the CIA. I suspect that the impact of using these alternative data sets would be to shuffle the points on the graphs without materially changing the conclusions but am unable to guarantee this.
PRESENTATION OF DATA
The Global Picture
Figure 1 compares per capita GDP in US$2013 against per capita electricity consumption in kWh/year for all 168 countries (used as a term of convenience; not all of them are independent countries). There is indeed a fairly strong relationship between the two (R2 = 0.79), but also considerable scatter. The relationship is also clearly stronger for the poorer developing countries than for the richer developed ones, and since wealth creation in the developing countries is the key requirement if the world is to eradicate poverty (a goal of the Kyoto Protocol) this analysis concentrates on the developing countries:
Figure 1: Per-capita electricity consumption vs. per-capita GDP, 168 countries.
Much of the scatter in Figure 1 is, however, caused by a few outliers that can legitimately be discarded and also by the European countries, which as we shall see behave somewhat differently to those elsewhere in the world. Figure 2 shows the results after the European countries and three outliers (Equatorial Guinea, Montserrat and the Virgin Islands) are deleted. The graph is now a lot cleaner, with R2 increasing to 0.84 and the trend line showing GDP increasing by $5 for every additional kWh of annual consumption:
Figure 2: Per-capita electricity consumption vs. per-capita GDP with European countries and outliers deleted
Figure 2 nevertheless still contains some outliers that are evident only when the data are segregated regionally. The next sections therefore segregate the global data into regions and geographic/economic groups:
Figure 3 plots the data for 53 countries in Africa, the poorest continent. The correlation is unimpressive (R2=0.32) but once more there are a number of outliers, including South Africa, which distinct from the other African countries already has a well-developed electricity sector, Libya, an oil-dependent country effectively in a state of war, Equatorial Guinea, where almost all of GDP comes from oil export revenues and other oil-dependent economies where GDP is similarly skewed by oil exports, such as Gabon, Angola and Nigeria:
Figure 3: Per-capita electricity consumption vs. per-capita GDP, 53 African countries
Deleting these outliers reduces the number of countries to 46 but yields much more consistent results, with R2 increasing to 0.90 (Figure 4). The inclusion of Mauritius and Seychelles, two island nations that have little in common with mainland Africa, is questionable, but R2 remains at 0.81 when they are omitted:
Figure 4: Per-capita electricity consumption vs. per-capita GDP, African countries with outliers deleted
According to the trend line GDP in Africa increases by about $4.50 for every additional kWh of annual consumption, close to the world average.
Figure 5 plots the data for 22 countries in Southeast Asia. The results are again not very encouraging, but they can be divided into five richer countries that show wide scatter and 17 poorer countries that do not:
Figure 5: Per-capita electricity consumption vs. per-capita GDP, 22 Southeast Asian countries
Two of the poorer countries on Figure 5 – Bhutan and the Maldives – are outliers; Bhutan because it generates a large amount of electricity from hydro plants funded by other countries and the Maldives possibly because of suspect data. Figure 6 plots the results for the 15 countries that remain after deleting Bhutan, the Maldives and the five richer countries. There is now another good correlation (R2=0.90), but in this case GDP increases only by about $2.50 for every additional kWh of annual electricity consumption:
Figure 6: Per-capita electricity consumption vs. per-capita GDP, Southeast Asian countries with outliers deleted
South and Central America
Figure 7 plots the data for the 18 countries in this data set, which contains no rich countries nor any grindingly poor ones and is therefore more consistent than the others. Unique among the data sets it also contains no obvious outliers:
Figure 7: Per-capita electricity consumption vs. per-capita GDP, 18 South and Central American countries
R2 in this case is 0.87 and GDP increases by about $5.00 for every additional kWh of annual electricity consumption.
Islands do not export or import electricity, are spread all over the world and range considerably in wealth. Figure 8 plots the data for 38 of the world’s islands (some of which are included in other categories). One of them – the Cayman Islands – is not representative of the general run of the world’s islands and three others (Montserrat, French Polynesia and the Virgin Islands) have potentially suspect data:
Figure 8: Per-capita electricity consumption vs. per-capita GDP, 38 island countries
Figure 9 replots the data for the 34 islands that remain after discarding these four outliers. R2 is now 0.89 and again each added kWh of annual electricity consumption corresponds to $5 of GDP growth:
Figure 9: Per-capita electricity consumption vs. per-capita GDP, island countries with outliers deleted
The data for 37 European countries are plotted in Figure 10. Iceland, Finland, Sweden and Norway are obvious outliers; Iceland and to lesser extent Norway because of the heavy industry that has gone there to take advantage of cheap hydropower and Sweden and Finland for reasons possibly related to the use of electricity in forestry industries and to the fact that they are downright cold in winter. Luxembourg, whose per-capita GDP is artificially inflated by the many capitas who work there but do not live there, is also non-representative. Switzerland is labeled for reference:
Figure 10: Per-capita electricity consumption vs. per-capita GDP, 37 European countries
Figure 11 plots the data for the 32 countries that remain after deleting Iceland, Finland, Sweden, Norway and Luxembourg. The clear distinction between the 16 Western European countries and the 16 Former East Bloc countries requires that the two be fitted with separate trend lines. But the correlations are modest at best (R2=0.35 for the western countries and 0.51 for the eastern), indicating that little confidence can be placed in the trend line gradients. Taken at face value they show each additional kWh corresponding to $7 of added GDP in Eastern Europe and $28 in Western Europe (although this number is almost certainly not meaningful. The Western Europe trend line gradient probably reflects the political and cultural differences that have become apparent during the ongoing Greek debt crisis as much as anything else.)
Figure 11: Per-capita electricity consumption vs. per-capita GDP, European countries with outliers deleted segregated into Eastern and Western subsets
The Oil Exporters:
Figure 12 plots the data for 18 countries whose economies are heavily dependent on oil exports. Two outliers, Qatar and (again) Equatorial Guinea, are excluded. The R2 value of 0.94 is the highest of any case considered and each kWh of consumption corresponds to about $4 of GDP. Clearly oil does not skew the link between wealth and electricity:
Figure 12: Per-capita electricity consumption vs. per-capita GDP, 18 oil exporting countries
That concludes the data presentation. The results show only a weak relationship between wealth and electricity in the developed nations, but R2 values of around 0.9 testify to the strength of the relationship in the developing nations. Except for Southeast Asia and Europe the trend lines also have gradients of $4 to $5 of GDP per kWh. (There is probably a message in this number but I have not had the time to look into what it might be. It costs a lot less than $5 to generate a kWh of electricity, so one intriguing possibility is that electricity might be acting as a wealth multiplier.)
So now on to the second question:
DOES ELECTRICITY CREATE WEALTH, OR DOES WEALTH CREATE ELECTRICITY?
The graphs presented above do not of course tell us which came first , so I address the question with specific country-by-country examples, again with the emphasis on the developing countries. It is impossible to review all of the 100-plus developing countries in the data base, but the examples presented below allow a conclusion to be reached:
1. Abundant cheap electricity does not necessarily create wealth:
Paraguay is a comparatively poor country that for years has had access to far more electricity than it knows what to do with. The electricity comes from Paraguay’s half share of the output from the massive 14GW Itaipu hydro plant on the Paraná River, which it owns jointly with Brazil. Itaipu began operation in 1984 and reached full production in 1991. Paraguay’s share of Itaipu’s output amounts to an average of about 44TWh a year, roughly four times its present consumption and almost twenty times its 1991 consumption. Has access to this abundant, cheap electricity make Paraguay richer? As shown in Figure 13, it has not. Annual GDP growth since Itaipu reached full production in 1991 has averaged only 1.5% and Paraguay’s constant-dollar per-capita GDP was still the same in 2005 as it was in 1991:
Figure 13: Paraguay’s per capita GDP, electricity consumption and electricity supply available from Itaipu
Why did Paraguay reap no benefit from Itaipu? Basically because it had nothing it could use the power for. Paraguay had no natural resource that it could exploit, nor was it a good place for power-intensive industries because Paraguay is a remote, landlocked country with no ocean access. So what did Paraguay do with the excess electricity? It “exported” it at cut-price rates to Brazil, although none of it touched Paraguayan territory on the way.
On other side of the globe is Bhutan, which also generates about five times as much hydropower as it can use (about 8 TWh/year versus 1.6 TWh annual consumption, according to my data). But this power has done little to enrich the country, which remains mired in poverty with a per-capita GDP on a par with Papua-New Guinea. And because the hydro is all run-of-river there is a massive surplus in the wet season and a deficit in the dry season which results in Bhutan being a net importer of electricity from India.
2. Wealth in the form of natural resources creates electricity:
New Caledonia is the second richest island in the island data set and also the largest per-capita electricity consumer (Figure 8). New Caledonia owes its wealth and high electricity consumption not to cheap and abundant electricity – 96% of it comes from fossil fuel plants that burn imported fuel – but to its enormous nickel reserves and to the large amounts of electricity that are consumed in mining and processing them. New Caledonia is a case where local wealth in the form of a major mineral resource created the electricity. And unless it can diversify its economy its wealth will last only as long as the nickel does.
The Seychelles has the third-highest level of per-capita electricity consumption in Africa and the highest per-capita GDP outside oil-rich Equatorial Guinea (Figure 3 and 4). The Seychelles along with a number of other islands enjoys a different type of wealth – beautiful beaches that attract tourists. And tourists want electric light, internet hookups, television and air conditioning, so if Seychelles wants the tourists to keep coming (they account for 70% of its GDP) electricity must be provided for them. Again the wealth creates the electricity.
3. Economic growth creates electricity in the absence of natural resources:
Mauritius is one of the world’s better-managed developing countries, and while its GDP is still relatively low it continues to grow at a healthy rate even though the island has no oil, gas, coal, minerals, hydro or well-developed tourist sector. Mauritius has achieved this by transitioning from an agriculturally-based to a diversified agricultural, manufacturing and financial services economy, and this transition has resulted in 8-9% annual growth in electricity demand. Mauritius is a particularly clear case of wealth creating electricity rather than the other way round.
4. Wealth and electricity sometimes go hand-in-hand:
Iceland, like Paraguay, also has far more cheap hydropower than it can use but unlike Paraguay has found a use for it – it sells it to aluminum companies that build energy-intensive smelters on the island. In this case the wealth-electricity relationship is symbiotic and it is impossible to say which created which. The refined aluminum would not be produced without the electricity, the electricity would not be generated without the aluminum and the smelters would never have been built in Iceland in the first place if the hydropower had not been there.
Did electricity create China’s wealth?
Figure 14 plots China’s annual electricity generation since 1985 (data from BP) against per-capita nominal GDP (data from the World Bank). The fact that the upturn in electricity generation around 2000 leads the upturn in GDP by about five years and tracks GDP thereafter suggests that the added electricity at least contributed to creating the added wealth, and it’s certainly hard to see how China could have become the world’s leading exporter without it:
Figure 14: Per capita GDP versus electricity generation, China
On the other hand China would not have had the electricity at all without its vast coal reserves, so once again the basic source of wealth was a natural resource that was ultimately converted via mines, power plants, imported raw materials, factories and ships into export earnings. Electricity was a vital link in the chain, but so were all the others. (An intriguing question is whether China’s wealth-creation approach would work in other countries. Would it have worked, for instance, in Botswana, which has arguably as much coal as China and which in the 1990s had a per-capita GPD approximately five times that of China? Probably not, if only because Botswana is another landlocked country with no seaport.)
The conclusion? In developing countries wealth creates electricity and not the other way round. There is no question, however, that once a country gains wealth it cannot sustain it without electricity. When the electricity disappears the wealth goes with it.