Journal of Legal Economics - Comment on "A Note on the Steady State Assumption and Expectancy Bias
November 28th, 2008McCollister and Pflaum (MP, 2007) maintain that “[f]orensic economists frequently treat dynamic variables considered in worklife and life expectancy calculations as fixed in the development of life and worklife tables. This procedure can lead to substantial errors.” Primary alleged offending authors are Richards (1999) in regard to occupation, smoking status, marital status, and language proficiency worklife tables; Ciecka and Goldman (1995) in regard to smoking; and Slesnick and Thornton (1997) for medical risks. MP focus on worklife by marital status in the one, and only, example calculation in their note. To estimate worklife expectancy, they “multiply the probabilities for marital status by those of the corresponding labor force participation status and the probability of living to that age.” That is, in order to capture the dynamic nature of marital status, MP make labor force participation a fundamental building block of worklife expectancy. Survival probabilities combined with labor force participation rates comprise the LP parts of the old LPE model. Skoog and Ciecka (2004) have shown that this model can be viewed as a Markov process, but it contains highly restrictive assumptions about transition probabilities between active and inactive states. With the usual notation (a for active, i for inactive, x for age, p for transition probability, pp for participation rate, and L for the number of survivors), the LPE model and MP implicitly assume that ^sup a^p^sup a^^sub x^ = ^sup i^p^sup a^^sub x^= (L^sub x+1^/L^sup x^ )pp^sub x^ ) and ^sup a^p^sup i^^sub x^=^sup i^p^sup i^^sub x^=(L^sub x-1^/L^sub b^)(1-pp^sub x^. In words, transitions are independent with respect to labor force state at age x. There are equal probabilities of being active at age x+1 whether active or inactive at age x, and equal probabilities of being inactive at age x+1 whether active or inactive at age x - assumptions decisively disconfirmed by estimates of transition probabilities by several authors (for example, see Krueger’s (2004) estimates of transition probabilities).
Posted in Legal Questions | No Comments »