The name is absent

Dewenter and Haucap (2005) provide a very promising approach for estimating termi-
nation rates. They use measures for market concentration and market size and control
variables for technology and regulation as well as country and year effects. I adopt this
approach and additionally include an investment parameter to get the following equation
for the analysis of the own investment effect on termination rates:

log(t_i¹_,l,z) = α_l^t_,¹_z + β_i^t_n¹_vlog(investmenti,l,z) + β_s^t1log(si,l,z) + β_m^t1_sizelog(msizel,z)

+β_u^t1_rbpoplog(share upopl,z)+regulation⁰_i,l,zβ_r^t1_eg+β₁^t1₈₀₀GSM 1800i,l,z+_i^t_,¹_l,z

where i is a firm index, l is a country index and z is a year index. Dewenter and Haucap
have also added the Herfindahl-Hirshman-Index (HHI) as a control variable and argue
that they have expected to find a significant impact of concentration on termination
rates. Nevertheless, the explanatory power of the HHI in their estimations might be
reduced due to the high correlation with the market share variables. Because of this
issue in the estimation specification and as it is found to have no significant effect on
termination rates I ignore it in the estimation approach. s is the individual market share
in terms of customers, msize is the total number of mobile subscribers in a country. The
share of urban population share upop is introduced as termination costs are expected
to depend on the population concentration. With a higher concentration termination
costs should be lower, thus, negatively affecting termination rates. Similar to Dewenter
and Haucap (2005) I introduce a GSM 1800 dummy to control for providers which only
offer communication via the higher, more expensive frequency level. As there is more
detailed information available about the alternative regulation schemes I add regulation
dummies for cost-based, incentive and asymmetric regulation, instead of the approach
provided in Dewenter and Haucap.

The equation for the amount of incoming traffic to the investor is specified as follows:

log       q_j¹_,i,l,z   = α_l^q_,¹_z + β_i^q_n¹_vlog(investmenti,l,z) + β_t^q1log(ti,l,z) + β_s^q1log(si,l,z)

j

+β_m^q1_sizelog(msizel,z) + β_p^q_o¹_stposti,l,z +β_u^q1_rbpoplog(shareupopl,z)   ⁽⁾

+regulation⁰_i,l,zβ_r^q_e¹_g + _i^q_,¹_l,z

15

<<   18   19   20   21   22   23   24   >>

More intriguing information