Abstract
Time is an important variable for retailers. The concepts of survival and duration, linked to this time variable, can be very interesting in franchising research. For instance, what are the determinants of the survival of a network or a store? Which elements decrease the period before franchising or internationalising a network? There is a well adapted but little-exploited methodology in this research area: survival analysis. Consequently, this methodological paper presents in detail survival analysis methodology before giving relevant examples of applications in the franchising field. Managerial implications of these kinds of research are given before the conclusion.
Notes
S(t) is a monotonic decreasing function such as S(0) = 1 and S(1) = 0.
Therefore, we have:
We can mention that the survival curve estimation by the actuarial method (Böhmer, Citation1912) is quite similar to the Kaplan-Meier methodology. Indeed, the Qj are estimated for time intervals a priori fastened and not time intervals determined by the dates of the events. In general, intervals chosen are the month, the semester, the year, etc. We notice that this actuarial method appeared before the Kaplan-Meier estimator.
We notice that when we compute Ŝ(t), only the dates at which some events are observed are taken into account. Ŝ is a constant between two times at which an event occurs.
Here, in this section, we only focus on time independent covariates.
c(β,Z ) depends on the features of the unit Z.
This ratio does not depend on time and these kinds of models are called proportional hazards models.