Estimating the Direct Rebound Effect in the Residential Energy Sector: An Application in Spain - Green Energy and Efficiency: An Economic Perspective - page 175

From the term electricity ( full ) log ( electricity price ) it is observed that house-

holds which use only electricity are more sensitive to variations in electricity price
than those that combine electricity with other fuels. Demand is inelastic for both
groups. Assuming the assumptions of symmetry, exogeneity and constant ef

×

ciency
are met, the direct rebound effect for households with more than one energy source
is 0.708, while for households that use only electricity is 0.866 (from the sum of

−

0.708 and 0.158). Substituting these gures in ( 6 ), a 10 % increase in energy

− −

ef ciency in heating and hot water systems entails electricity savings of 2.92 % in

the rst group of households, and 1.34 % in the second group.

4.2 Demand for Natural Gas

Table 6 shows the results of the Heckman model applied to the demand for natural
gas. The panel (a) shows the coef cients of the consumption model and the panel (b)

contains the results of the probit model whose dependent variable represents the use
of natural gas. 6 Natural gas heating is hardly used at all in the households in
the southern area, so these households are not included in the natural gas model. This
reduces the size of the sample to 474 households, 271 of which use natural gas.
The parameter ρ , which represents the correlation between unobserved errors in
the selection and demand equations, is signi

nding validates the
use of the Heckman model for analysing natural gas demand. Based on the selection
model, it is deduced that electricity and natural gas prices are not signi

cant at 1 %. This

cant in the
choice of natural gas. The variables surface area of dwelling, rented dwelling, CO 2
emissions and household in northern area are found to be signi

cant at 1 %. The

rst three of these variables negatively affect the choice of natural gas, while the last
increases the likelihood of its use.
The variable number of rooms is found to be signi cant at 5 %, and to increase

the likelihood of natural gas being chosen as a residential fuel. The remaining
variables,

2,500/month, household with electric oven and
household with dishwasher are signi

i.e. income € 1,500

−

cant at 10 % and have a positive effect on the

choice of natural gas.
Regarding natural gas consumption, the variable surface area of dwelling, which
can be interpreted as elasticity, is signi cant at the 1 % level. Hence, a 10 % increase

in the household surface area would increase natural gas consumption by 13.4 %.
Income related variables are signi

cant at least at 10 %. These coef cients

indicate that households where the income is more than 1,500/month consume

approximately 30 % less natural gas than those with lower incomes.

€

6 For the Heckman model to perform well it is recommendable that at least one variable that only
affects selection (in this case of natural gas) should be signi cant [ 12 ]. This paper assumes that
owning more electrical appliances may provide an incentive to use electricity along with other
energy sources such as natural gas.