## Energy Economic Example Energy Intensity of Consumer Goods and Services Energy Cost of Living

This topic is included to emphasize and illustrate the breadth of applicability of energy analysis and the analogies between ecological and economic systems. The question is how much energy is required to support, directly and indirectly, human household consumption patterns. The approach is in two steps: (1) determine how much energy is needed, directly and indirectly, to produce a product; (2) determine how much of it a household consumes.

Consider a loaf of bread. The energy to grow the ingredients, make the bread, and transport and market it can be determined by a detailed vertical analysis (also called process analysis), in which one sums:

1. the energy used in the supermarket;

2. the energy consumed in the bakery;

3. the energy consumed at the flour mill;

4. the energy used on the farm;

5. the energy for transport at every link; and so on.

This process can even lead to cycles in systems with feedback (e.g., cars require steel, but the steel industry uses some cars), but the process usually converges to an acceptable answer after just several steps.

A vertical analysis is potentially accurate, but expensive. Performing it for a wide range of products is prohibitive. There is, however, a large database on the interactions of the sectors (c. 350-500) of the US economy. This is the input-output (I-O) table published by the US Department of Commerce. Many other countries have similar I-O tables. With a number of fairly stringent assumptions, this table can be combined with direct energy use data for each sector to produce energy intensities using the equation implied by Figure 3b. One such assumption is necessitated by the fact that the units in I-O tables are monetary units per year, so one must accept dollars as an appropriate allocator of embodied energy. Because in the American energy industry, energy is usually measured in Btu, the energy intensity of goods and services is then expressed in Btu/\$. I-O-based determination of energy intensities has been performed for c. 35 years. Under further assumptions, the intensities can be used to evaluate the energy impact of different expenditure patterns. Doing this for a household yields the so-called energy cost of living.

The I-O data are available, but gathering the associated direct energy data and performing the computation is tedious, though today's computers make it increasingly easier. Solving 500 simultaneous equations of the form in eqn [1] is done by inverting a 500-rank matrix.

Once we have the energy intensities, we need details on how households spend their money over the range of consumer product categories, also known as their market basket. This information is collected by the US Bureau of Labor Statistics. Putting the two together yields the energy cost of living (ECOL):

i=all expenditure categories where Yi is the household's annual expenditure for expenditure category i. Applying eqn [6] allows one to analyze the effect of overall spending and the mix in the market basket. The latter will be significant only if the energy intensities are different for different expenditure categories.

Table 5 shows I-O-based energy intensities determined by Carnegie Mellon University for 1997, and updated and aggregated by the author into 15 categories covering all household expenditures. The intensities are indeed different, especially energy itself and service industries such as health care.

Figure 14 shows the result of transforming of a household market basket (in dollars/yr) to its energy impact (in Btu/yr) using eqn [6] and intensities such as those shown in Table 5. In Figure 14a, we see that of the average household's expenditures of \$49 300 in 1973, only 6.4% was for direct energy (residential fuel and electricity and auto fuel). After conversion to energy requirements (Figure 14b), this portion was 63% of the total impact of 604 million Btu. The total is roughly the energy equivalent of 100 barrels of oil. Figures 15a and 15b show the energy

Table 5 Energy intensities for household consumption categories (Btu/\$, 2003 technology, 2003 dollars)

Table 5 Energy intensities for household consumption categories (Btu/\$, 2003 technology, 2003 dollars)

 1. Residential fuel, electricity 139300 2. Vehicle fuel 94300 3. Vehicle purchase, maintenance 5 400 4. Food 6100 5. Alcohol, tobacco 3 700 6. Apparel 6500 7. Communication, entertainment 4000 8. Health, personal care 2 400 9. Reading, education 3000 10. Insurance, pension 1 600 11. Contributions 3800 12. Public transportation 21 200 13. Asset gain 4 700 14. Miscellaneous 4200 15. Housing 5100 Direct energy ((1) + (2)) 118100 Nonenergy (sum of (3)-(15)) 4 700 All personal consumption 11 100 Energy/GDP 8900 Sprawl ((1) + (2) + (3) + (15)) 15 700 Nonsprawl (sum of (4)-(14)) 4300 Auto and related ((2) + (3)) 21 200

Shaded categories indicate an intensity greaterthan the energy/GDP ratio. 'Sprawl' contains housing and auto ownership and operation. Source: Author's calculations based on Carnegie Mellon University data.

Shaded categories indicate an intensity greaterthan the energy/GDP ratio. 'Sprawl' contains housing and auto ownership and operation. Source: Author's calculations based on Carnegie Mellon University data.

pie for the lowest expenditure decile (\$11 500/yr, 241 million Btu/yr) and highest decile (\$140 200/yr, 1233 million Btu/yr). The direct fraction is largest, 79%, for the lowest decile, and lowest, 47%, for the highest decile.

Figure 16 shows a statistical fit to energy versus expenditures for a representative sample of several thousand American households. It confirms that because the mix changes, energy is not a linear function of total expenditures, but rather bends down and away from a straight line through the origin. The reason is that direct energy (auto fuel and residential fuel and electricity) tends to level out as household expenditures increase. Expenditures increase for other products, but these tend to be less energy intensive. For developed countries, the shape of Figure 16 seems robust: studies of Norway, the Netherlands, and Australia have found a similar result.