Show that the duration data model with Pareto density of the first kind $f(t)=\alpha k^{\alpha }/t^{\alpha +1},\alpha >0,t⩾k⩾0$, is an accelerated failure time duration model but is not a proportional hazards model. [Show that In $t$ can be expressed as a linear regression in $k=\exp \left({\mathbf{x}}^{\mathrm{\prime }}\mathit{\beta }\right)$ with an additive homoskedastic error.]