Appendix
In this appendix we derive the likelihood function for the empirical model described by equations (1) and
(2) in the text. For simplicity of notation, we drop from the notation the dependence of functions on the
predictor variables z and x and, where appropriate, Tf. Let Pf(kf∣θf) denote the probability ofenrolling in
college at time kf conditional on θf and Pfc(kf]θf) equal the probability ofthe spell lasting longer than kf.
Then,
Pf( kf | θf ) = Sf( kf | θf ) - Sf(kf+1 | θf ), (A1)
Pfc( kf | θf ) = Sf( kf | θf ). (A2)
In a manner similar to McCall (1996, p. 679), let Pg (ks | θg, θd), Pd(ks | θg, θd) and Psc(ks ^g, θd) be the
probability ofgraduation after ks years ofenrollment; dropping out after ks years ofenrollment; and right
censored after ks years of enrollment, respectively. Thus,
Pg( ks | θg, θd ) = S( ks ,ks| θg, θd )-S( ks+1,ks| θg, θd )-.5 [S( ks ,ks| θg, θd )+S( ks +1,ks+1| θg, θd)
-S( ks ,ks+1| θg, θd )-S( ks+1,ks| θg, θd )], (A3)
Pd( ks | θg, θd ) = S( ks ,ks| θg, θd )-S( ks ,ks+1| θg, θd )-.5 [S( ks ,ks| θg, θd )+S( ks +1,ks+1| θg, θd)
-S( ks ,ks+1| θg, θd )-S( ks+1,ks| θg, θd )], (A4)
and
Psc( ks | θg, θd ) = S( ks ,ks| θg, θd ) (A5)
where the term
28
More intriguing information
1. Skills, Partnerships and Tenancy in Sri Lankan Rice Farms2. The name is absent
3. Linkages between research, scholarship and teaching in universities in China
4. Why unwinding preferences is not the same as liberalisation: the case of sugar
5. Uncertain Productivity Growth and the Choice between FDI and Export
6. The name is absent
7. Demand Potential for Goat Meat in Southern States: Empirical Evidence from a Multi-State Goat Meat Consumer Survey
8. SOME ISSUES IN LAND TENURE, OWNERSHIP AND CONTROL IN DISPERSED VS. CONCENTRATED AGRICULTURE
9. Imitation in location choice
10. The name is absent