A Method for Analyzing Multiple Spell Duration Data

摘要

In recent years many interesting papers in economics have been written on the general subject of analyzing duration date. This is especially true in the fields of human fertility and labor economics. For example, Flinn and Heckman (1981), Heckman and Borjas (1980), Kiefer and Neumann (1981), and Lancaster and Nickell (1980) study either employment or unemployment duration data. In similar papers, Mincer and Polachek (1974, 1978) and Mincer and Ofek (1980) use regression analysis to study unemployment duration data on married women. Gomez (1980) studies fertility duration data and Harris (1980) looks at infant mortality duration data. Fenn (1981) analyzes duration data on sickness absence and Kennan (1981) and Lancaster (1972) study strike duration data. In all of these papers the data are of the single spell type. That is, the analyst examines only one spell of history for each individual in the example. The case of multiple spell data is more difficult, but two papers that look at the problem are Chamberlain (1981) and Heckman (1981b).