臺灣失業勞工尋職行爲之研究－區分母體存活模型之應用

本文提出以傳統存活分析(survival analysis)研究失業勞工尋職行爲時常忽略的一項問題，即：失業勞工群中存在部分比例的失業勞工，由於主觀或客觀條件因素使然，造成其永遠無法由無工作狀態轉進至就業狀態，若循傳統的「單一母體存活模型」(single population survivor model)分析，而忽略了此一現象，則失業期間之分配將被扭曲，進而錯估共變數對就業轉機率（或失業期間）的影響效果。此外，因理想資料的匱乏，常使得研究者無法將存活模型由「單項終點」進化至「多項重點」(multiple destinations)，並以「競爭性風險模型」(competing risks model)加以分析。本文應用「區分母體存活模型」(split population survivor model)，不將退出勞動市場的選擇視爲失業勞工結束失業狀態的一項終點，而將其視爲影響失業勞工轉進至就業這個單項終點狀態的一個因數，如此將達到與競爭性風險模型有異曲同工之估計結果。

關鍵字

區分母體存活模型；就業轉機率；勞動參與；失業期間；異質性

並列摘要

The thesis proposes to study an issue that is often overlooked in the job searching behaviors among unemployed workers by utilizing the conventional survival analysis, which pertains to how a certain percentage of unemployed workers among the unemployed workers groups, as curtailed either by objective or subjective conditional factors, would not be able to advance from a non-working status to an employed status. Of analysis done by the conventional single population survivor model, ignoring the phenomenon could poise to distort the distribution of the unemployment duration, which in turn tends to lead to erroneous estimation on how covariates come to affect the end results of the employment hazard rate (or the unemployment duration). Moreover, the lack of ideal data that often prevents researchers from advancing the survivor analysis model from a single destination to a set of multiple destinations, or successfully analyzing the data using the competing risks model. The thesis proposes to adopt the split population survivor model that regards the choice of withdrawing from the labor market not as a destination for unemployed workers to conclude their unemployed status, but rather regard it as a factor that hinders unemployed workers from progressing to the single destination status of switching to being employed, which essentially provides comparatively similar estimation results as would by the competing risks model.