The Unassuming Economist

From VoxEU:

With global emissions of CO₂ still growing, understanding the determinants behind energy use and emissions is as relevant as ever. This column looks at the role of per capita income and consumption choices. It finds that the share of expenditures spent on energy and energy-intensive goods tends to decrease with income across a large set of countries. Simulations indicate that income growth shifts consumption patterns in a way which generally reduces emissions. However, increasing emissions in low- and middle-income countries as well as a shift from direct to indirect consumption of energy mean that the effect on total world emissions is modest.

The consumption of energy is associated with numerous negative externalities. Fossil fuels, specifically, cause emissions of CO₂, the most prevalent greenhouse gas. With the issue of climate change far from resolved, understanding the determinants behind emission levels is necessary to improve forecasting and guide policy making.

There is ample empirical evidence that energy demand depends on per capita income. In high-income countries, a number of single-country survey-based studies find that rich households spend a smaller share of their income on energy (e.g. Dubin and McFadden 1984). Less is known about this relationship in the developing world, where most of the growth in energy demand is currently taking place. A number of studies find that energy demand also increases slower than total expenditures in middle-income countries, including in China (e.g. Cao et al. 2016). On the other hand, as large populations are just beginning to purchase energy-intensive appliances (fridges, air conditioners, cars), it is feared that energy demand in the developing world may keep growing rapidly (Auffhammer and Wolfram 2014, Gertler et al. 2016).

More generally, the structural change literature has clearly documented that consumers of different income levels prefer different sets of goods (e.g. Comin et al. 2018), generally shifting from agriculture to manufacturing to services as income grows. Yet we still do not have a clear picture of the aggregate consequences for energy consumption, and what the relationship between income, consumption patterns and CO2 emissions looks like in the aggregate. This relationship directly affects the CO₂ intensity of GDP, a critical ingredient when trying to project emissions and understand the potential magnitude of climate change. A quick survey of prominent emission projection models (from academics, governments, or international agencies) also shows that these models tend to incorporate income effects in consumption in a rather crude manner, as they are typically more focused on technology assumptions.

Our recent article (Caron and Fally 2018) systematically summarises the implications of shifting consumption patterns on energy demand and emissions. We take an aggregate view across countries spanning most of the per capita income spectrum and all sectors of the economy. Our framework, an extension of the general equilibrium model described in Caron et al. (2014), links non-homothetic household preferences (i.e. with non-trivial income effects) to emissions through production and trade linkages. The model allows us to uncover patters of comparative advan­tage in the production of CO₂-intensive goods. We can thus identify the role of income in determining energy demand, making sure that we are not confound­ing it for differences in the availability of energy or energy-intensive goods and services across countries.

The relationship between income elasticity and emissions intensity across sectors

Do high income consumers prefer goods which are, coincidentally, less CO₂ intensive in their production?

Using different types non-homothetic preferences, we estimate the income elasticity of goods in 50 sectors of the economy. As many before us, we find a clear role for per capita income in determining consumption patterns – the income elasticity of most sectors is significantly different from one.

The directconsumption of energy generally increases slower than income. Figure 1 shows that the two main energy goods consumed directly by households, electricity (ely) and refined oil (p_c, mostly gasoline for transportation), have an income elasticity below one. While these estimates reflect an average across all countries, they depend on income – energy consumption is more income-elastic in low-income countries. This is consistent with scattered country-level evidence from the literature.

However, just looking at direct energy consumption is misleading. It ignores the emissions embodied in the consumption of non-energy goods – think of the emissions caused by the production of the steel and leather constituting the chair you are sitting on, and so on. Tracking ‘indirect’ emissions throughout global supply chains (based on intermediate input and trade linkages), we can compute the ‘total’ emissions associated with the consumption of each good. Plotting income elasticity against total (embodied) CO₂ intensity, Figure 1 reveals an ‘inverted-U’ relationship – sectors of intermediate income elasticity have, on average, the highest CO₂ intensity. But it is also asymmetric and negative overall, with high-income-elasticity sectors having the lowest CO₂ requirements on average.

We insist that this represents an ‘incidental’ correlation – high-income consumers do not necessarily prefer these goods becausethey are less emissions intensive.”

 

Continue reading here.