These results show that tighter noise regulation under a cumulative constraint hurts airline
passengers, with the impact operating through two channels. Flight frequency falls, reducing
service quality, and fares rise, so that the cost of travel escalates as it convenience declines.
For tighter regulation to be desirable, these losses must be offset by social gains from a lower
noise level, as analyzed below.14
3.2. Sequential choice under a cumulative constraint
Using the simultaneous-choice case as a benchmark, consider now the case where frequencies
and fares are chosen sequentially. Under sequential choice, fares are set conditional on
frequencies, with frequencies then chosen in a first stage, taking into account the second-stage
impact on fares.
To analyze this case, the first-order condition for choice of p1 (eq. (9)) is supplemented
with the analogous condition for airline 2, gotten by reversing the 1 and 2 subscripts in (9).
The equations are then solved for p1 and p2 as functions of f1 and f2. The solution for p1 is
given by
p1 = α/2 + τ − 1
3
γ
f1
− γ
f2
+
2f1 + f2
L
, (14)
and the p2 solution is gotten by reversing the 1 and 2 subscripts. These solutions are substituted
into the profit function (8) for airline 1, which is differentiated with respect to f1 treating f2 as
parametric. Symmetry is imposed in the resulting first-order condition, which is then solved
for f.15 The result is
f =
γ
3θ + /L
, (15)
which differs from (12) by the presence of the factor 3, rather than 2, in the denominator.
Thus, for a given L, frequency is smaller in the sequential-choice case. The p solution in (11),
which remains relevant, shows that the smaller f leads to a lower fare under sequential choice.
In addition, the lower f leads to a smaller aircraft size, and aircraft are also noisier under
sequential choice (the 2 factor in (13) is replaced by 3).
Despite these differences, the timing of airline decisions has no effect on the main
comparative-static properties of the equilibrium. In particular, it is easy to see that the change
12
in the f solution leaves the signs in Table 1 unaffected, so that the table applies to both the
sequential and simultaneous-choice cases.16
The key conclusions of the preceding analysis are then summarized as follows:
Proposition 1. A reduction in the allowed noise level under a cumulative noise
constraint leads to lower flight frequency, larger and quieter aircraft, and a higher fare,
regardless of whether airline choices are simultaneous or sequential. Airline passengers
are hurt by the tighter noise limit.
3.3. The case of a per-aircraft noise constraint
Suppose now that noise regulation takes the form of a per-aircraft noise constraint, written
as ni ≤ n, i = 1, 2. Since the constraint will bind, airline 1’s objective function is given by (8)
with n1 replaced by n. Under simultaneous choice, the previous first-order condition (9) for p1
remains relevant, but with L replaced by f1n. Imposing symmetry, the fare solution is then
p = α/2 + τ + /n, (16)
which corresponds to the previous solution in (11) after substitution for L. This equivalence
implies that, if L under the cumulative constraint is set so that noise per aircraft in (13) is
equal to n, the same fare levels emerge under the two regimes.
The first-order condition for f1 is now
∂π1
∂f1
=
γ(p1 − τ − /n)
αf2
1
− θ = 0, (17)
and, substituting (16), it yields a symmetric frequency solution of
f =
γ
2θ
. (18)
Under sequential choice, which is analyzed by following the same steps as before, the fare
solution in (16) remains relevant but frequency is given by
f =
γ
3θ
, (19)
13
where a factor of 3 replaces the 2 in the simultaneous solution, as in the cumulative case.17
Under both the simultaneous and sequential solutions, frequency is increasing in γ and
decreasing in θ, as in the cumulative case. But frequency does not depend on n and is thus
independent of the stringency of the noise constraint, in contrast to the cumulative case. With
f independent of the noise limit, nf and thus total noise L rises as n is raised, and L rises
with γ and falls with θ, holding n fixed. In addition, since the fare solution in (16) does not
involve f, p is now independent of γ and θ, although the fare responds positively to α and
τ , as before. These comparative-static effects are summarized in Table 2. Airline passengers
are hurt, as in the cumulative case, by a tighter noise constraint, but the effect now operates
through a single channel, a higher fare, with service quality unchanged.18
As in the cumulative case, sequential choice yields a higher frequency and smaller aircraft
size than simultaneous choice. But since p is independent of f, fare levels are the same under
the two choice scenarios, in contrast to the previous conclusion.
A more interesting observation, however, is that frequency is always higher under a peraircraft
noise constraint than in the cumulative case, with the opposite conclusion applying to
aircraft size. These results follow because the simultaneous frequency solution in (18) is larger
than the corresponding cumulative solution (12), with the same comparison applying to the
sequential solutions (19) and (15). Apparently, lower frequencies arise in the cumulative case
because a reduction in f provides a means of satisfying the cumulative constraint.
Summarizing the main conclusions of the above analysis yields
Proposition 2. (i) A reduction in the allowed noise level under a per-aircraft constraint
raises the fare, harming airline passengers, while having no effect on flight
frequency or aircraft size. Total noise falls.
(ii) Flight frequency is always higher and aircraft size lower under a per-aircraft noise
constraint than under a cumulative constraint, but fares are identical in the two cases
when the constraints are set to achieve the same noise per aircraft.
3.4. The monopoly case
In the monopoly case, analyzed by Girvin (2006a), different comparative-static results
emerge. Although closed-form solutions for the decision variables are not available, her analysis
14
shows that a tightening of the cumulative noise constraint reduces flight frequency, as in
Proposition 1. A lower L also affects traffic q, which no longer assumes the constant equilibrium
value of 1/2, causing it to drop. With f and q both falling, the change in aircraft size is
ambiguous, and the same conclusion applies to aircraft quietness and the fare. Under a peraircraft
noise constraint, a reduction in n lowers f, in contrast to Proposition 2, while also
reducing traffic. In addition, aircraft size falls. These conclusions show that the duopoly and
monopoly equilibria have very different properties.