What is going on? Why does a seemingly natural way of modelling (de)merit
good arguments lead to results whose intuitive appeal depends on the elasticity
of substitution? Let us consider the limiting case of Leontief preferences, and
ignore—for the sake of graphical representation—the standard non-numéraire good.
Suppose that a person’s preferences can be represented by min{z, y}.Thisperson
has L-shaped indifference curves with the corners lying on the 45o line, as shown
in figure 1.
Leontief preferences and a demerit good
The government, on the other hand, thinks of good y as a demerit good and
subscribes to the preference ordering represented by min{z, 1 y}. The associated
indifference curves are the dashed lines. Clearly, the government’s preferences
are more favourable to commodity y than the agent’s preferences are!
With Leontief preferences, no finite subsidy will be able to distort the agent’s
budget allocation away from the laissez-faire solution, but once the degree of sub-
stitutability becomes positive, it will. The reason why discounting a commodity
leads to subsidisation should now be clear. The government respects that red
wine is complementary to a pasta meal. But it regards one bottle of wine only
half as good as you do. In order to get the maximal utility out of a pasta dish,
it wants you to drink more wine, not less.
This paradox holds true more generally. Notice that the government’s MWP
for the (de)merit good (θЦ3(z, x, θy)) has the following elasticity w.r.t. θ:
dlog( 3 ) = 1 - (_∂log y(z,x, θy)
∂logθ I ∂ log(θy)
(10)
Thus, whether the government’s MWP exceeds or falls short of that of the con-
sumer depends on whether the (own) demand price elasticity for the (de)merit
good exceeds or falls short of 1 in absolute value. If the demand for tobacco, say,
is inelastic, the demand price elasticity is likely to be large, and under the scaling