Gerhard Tintner and Gabriele Wörgötter, The Demand for Energy in Private Households in Austria (1976)

Tintner and Wörgötter’s conference contribution examines Austrian household energy demand through separate estimates for coal, electricity, natural gas, heating oil, and petrol. Its central finding is that aggregate income and prices affect these energy sources differently: coal consumption declines with rising income, electricity and oil consumption expand with household equipment, and petrol demand depends strongly on the number and engine capacity of cars. The contribution proceeds from fuel-specific regressions and comparative conclusions to a structural model of petrol consumption. Household heating and electricity estimates use annual observations for 1958–72; the petrol model covers 1956–72.

The analysis is explicitly exploratory. The authors report only equations yielding plausible results and warn that autocorrelation, multicollinearity, and simultaneous relationships may bias their estimates. Their methodological qualification governs how the numerical findings should be read:

Most estimates are provisional and have to be refined by further investigations.

All variables are transformed into logarithms, allowing regression coefficients to be interpreted as elasticities. The reported standard errors, multiple correlation coefficients, and Durbin–Watson statistics accompany the substantive interpretations, but the authors do not treat apparently strong statistical fit as sufficient evidence of a satisfactory explanation. Their account repeatedly turns to household circumstances and changes in energy-using equipment to explain unusually large coefficients.

Coal provides the clearest connection between an estimated demand relationship and a social interpretation. Its income elasticity is approximately −0.79, its own-price elasticity −2.73, and its temperature-index elasticity 1.27. The authors identify the negative income effect directly:

Income-elasticity is negative, that is coal is an inferior good.

They associate coal with poorer households, suggesting that these consumers reduce consumption sharply when prices rise. Coal is also the only fuel for which they identify a significant response to temperature changes. The result distinguishes it from other energy sources within the reported estimates; it does not establish that weather is generally irrelevant to household heating. Their conclusions favor simple coal-demand equations and identify natural gas as the other fuel whose price merits consideration.

Electricity presents the opposite income pattern. Its estimated income elasticity is 1.847, while its own-price elasticity is −0.663. The authors interpret the more-than-proportional income response through the growing stock of electrical equipment:

(The increasing amount of electric equipment raises the consumption of electricity more than proportional to income.)

The explanatory move is important: income does not merely finance greater use of existing appliances but accompanies changes in the equipment through which energy is consumed. The conclusions again favor parsimonious equations, while noting that heating-oil and coal prices matter more than natural-gas prices in the alternative specifications considered.

Natural gas produces the least secure findings. Although the estimated coefficients have the expected signs, some are insignificant. More fundamentally, the authors attribute bias to the public gas company’s increasing use of natural gas, indicating that a changing supply context complicates the interpretation of demand. They propose introducing a dummy variable rather than accepting the reported regression as definitive:

For all these reasons one cannot give definite conclusions, but at least it is possible to say that other prices have a large influence on the demand for natural gas.

Heating oil has an exceptionally high estimated income elasticity of 4.402 and an own-price elasticity of approximately −1.2. The authors consider the price response reliable but regard the income response as excessive despite its statistical significance. Their explanation again invokes equipment: oil-fired central heating was expanding among households wealthy enough to afford it. The coefficient may therefore capture the spread of a heating system as well as consumption within existing installations. The displayed equation’s positive natural-gas cross-price coefficient is insignificant; the broader conclusions instead emphasize coal prices and caution that their estimated cross-price effect may also be too large.

The petrol section makes the equipment-mediated argument explicit through a structural model. Income helps determine the number of cars; average engine capacity depends on its lagged value, petrol prices, and income; petrol consumption depends on income, price, vehicle numbers, and engine capacity. Reduced-form relationships are then used to recover a consumption equation by indirect least squares. This distinguishes immediate demand effects from effects transmitted through the vehicle stock.

The resulting direct income elasticity is only 0.06, compared with approximately 0.713 for vehicle numbers and 2.179 for average engine capacity. The authors explain why the small income coefficient should not be mistaken for an absence of income effects:

Rising income first increases the number of vehicles, and, with an increase in the number of cylinders (the DHR), the consumption of petrol.

With vehicle numbers and engine capacity held constant, income contributes little additional explanatory power. The low petrol-price elasticity, approximately −0.141, is interpreted as evidence that petrol is a necessity. The contribution’s broader relevance lies in this distinction between direct responses and changes mediated by durable equipment. Across fuels, the estimates connect household income, social differentiation, and technological adoption, while the authors’ repeated qualifications preserve the difference between plausible economic interpretation and established causal explanation.