The first post in this run looked at the crude data showing the strong link between energy use and wealth.
But a relationship by itself doesn’t indicate causality. The second post presented evidence suggesting that the relationship doesn’t run from wealth to energy use, and may well run the other way.
The data from the first post also suggested an increase in the energy efficiency of economies over time, which is the subject of this post. (For the sources of the data in all these posts, see here.)
As a reminder, we measure the energy efficiency of an economy by the energy used per unit of GDP (in this case, it will be in terms of tonnes of oil equivalent [toe] per $1,000 of GDP. A smaller number means fewer toe per $1,000 GDP, which means a more efficient economy.
One hypothesis is that countries get more efficient as they get wealthier, and there are a few reasons to think that might be true.
- A richer country can afford to build more efficient factories, more efficient houses, more efficient cars.
- As a country develops, its service sector tends to grow faster than its industrial sector. Services use less energy per worker than industry, so that shift will reduce the country’s energy per GDP.
- A piece of why the industrial sector shrinks (in relative terms) is that a more developed country may start to import a greater quantity of the goods it uses (steel, cars, TV’s, computers, etc.), and pay for that with exports of services (movies, banking, tourism, education, etc.). So the richer countries have taken some of the production of things they’re going to consume and moved it to poorer countries.
One way to examine this hypothesis is through a cross section, looking at our sample of countries in a given year. Each mark on the chart represents a country, and its wealth (GDP per capita) is shown by horizontal position, while its efficiency (tonnes of oil equivalent, or toe, per million dollars of GDP) is reflected by its vertical position.
Since using less energy per GDP represents greater efficiency, our hypothesis suggests that countries should be spread across the chart roughly as shown in Figure 1, with richer countries (far to the right) having higher efficiency (being relatively low), while poorer countries (at the left) don’t have good efficiency (so they’re positioned high on the chart).
|Figure 1. Hypothesized relationship between wealth and efficiency|
Figure 2 shows a cross section for 1970, and ... it’s not clear.
|Figure 2. Data from BP Statistical Review and Penn World Tables|
If you look at the “frontier” (the rough line of points furthest from the origin, the bottom left corner of the chart), they do follow the expected shape (as highlighted in Figure 2a).
|Figure 2a. Data from BP Statistical Review and Penn World Tables|
But there’s a mass of points that seem to go in exactly the opposite direction (as highlighted in Figure 2b), rising up to the right from the origin. For this group of countries, efficiency gets worse as wealth increases.
|Figure 2b. Data from BP Statistical Review and Penn World Tables|
Figure 3 gives the cross section for 1990. Still a very mixed message.
|Figure 3. Data from BP Statistical Review and Penn World Tables|
Figure 4 shows 2014, and now the shape is noticeably different. It seems to move left-right, without any clear trend up or down.
|Figure 4. Data from BP Statistical Review and Penn World Tables|
(Figure 4 has been truncated vertically axis, to exclude the one very odd point of Trinidad and Tobago, which shows up in the data as a middle-income country with bizarrely bad energy efficiency. And it’s been truncated horizontally to exclude a few small, very rich, oil-exporting countries. Those truncations make it easier to see the shape among the great majority of countries in the dataset.)
Figure 5 overlays all three years: 1970, 1990, and 2014.
|Figure 5. Data from BP Statistical Review and Penn World Tables|
Going from 1970 to 1990, (black circles to orange squares), the least efficient countries in 1990 aren’t as bad as the least efficient countries in 1970 (fewer extremely high points). And there are a lot more fairly efficient countries (more orange squares that are relatively far down on the chart).
It does seem, however, that even though the extreme inefficiency isn’t as bad in 1990, there are more countries that are fairly inefficient. But that’s largely because the 1990 sample has countries that weren’t in the 1970 data, and these are generally Soviet Republics (e.g., Russia, Lithuania) and recently post-communist countries (e.g., Poland, Hungary), and these tended to have pretty poor energy efficiency.
But if we compare those first two years to 2014 (blue circles), there’s a marked change. The mass of data points has generally moved down and to the right, as the countries in the data have, in aggregate, gotten wealthier (to the right) and more efficient (down) over time.
Figure 5 also suggests another interesting possibility, in that there seems to be sort of an “envelope” at 50 toe per million $ GDP. A few poor countries in 1970 and 1990 are just below that level, and then by 1990 a handful of poor, middle-income, and rich countries are just above it (see figure 5a).
|Figure 5a. Data from BP Statistical Review and Penn World Tables|
Is this some sort of barrier, some maximum level of energy efficiency that can’t be breached?
This picture doesn’t give us enough information to say that.
Really, nothing can tell us that except future data—if we go a long time without breaching it, then there’s reason to think it may be a limit.
The next post will explore what we can learn about efficiency changes over time, by looking at time paths of individual countries.