change is the sum of the cost-related quantity change and the termination rate-related
quantity change. Moreover, we know from the derivative of own termination rates to
investments that the change in termination rates is half the change in per-unit costs.
Thus, equation (17) changes to:
∂πi i k
j,i i-Z
dki πj,i
∂ti ki + d Pj qj,i ti ∂ti ki
dki ti dti PP j qj,i dki ti
βt1 + βq1βt1
- inv + t1 inv
(18)
Similarly, one has to add the significant regulation coefficients of table 5 for the in-
vestment effect on own profits.18 Doubling investments increases short-run profits from
incoming traffic by about 4.7 to 6.8 percentage points (when including interaction terms).
Similarly, the relative investment externality on competitors’ profits is calculated. The
relative profit change due to a one-percent change in investments is given by:
∂Pj ∏-,,i ki = ∂j ki + ∂ P-j q-ij ki
(19)
(20)
dki P-, ∏-lj dki tj dki P-, q-j,j
Rewriting equation (19) yields:
dP-j πjj,j ki ∂t ∂tj ki , dP-j q-j,j c0(k ) ki , dP-j q-j,j tj ∂tj ki
dki P-j ∏Lj,j dki tj' dci i^ i' P-j q-j,j ^+^ dtj P-j q-j,j dki tj
= βt2 , βq2 , βq2βt2
inv inv t inv
By replacing the coefficients with the estimation results in tables 4 and 5 we find that
the competitors’ wholesale profit changes by -1.2 and -2 percentage points, respectively.
While there exists a strong impact of investments on termination rates and also on traffic
the impact on profits is close to nil. In contrast to no investment effect on the investor’s
own incoming traffic the increase in incoming traffic to competitors is relatively large
and, thus, reduces the negative effect on wholesale prices in the competitors’ profit
function. Thus, at least in the short-run the effect on incoming traffic cannot outweigh
the reduction in termination rates.
The question still remains why competitors reduce termination rates if they are not
obliged to do so. One admittedly speculative consideration might be the following: The
players in a market know the competition-driving variables of the market which are in
particular prices (on the retail and the wholesale level) and traffic. Moreover, as they are
18I only consider the common estimation approach in columns (3) as this is the correspondent estimation
approach to table 4.
27