Premium
Empirical Analyses of Causal Relationships among Estimated Aggregate Unemployment Duration and Some Properly Chosen Economic Variables
Author(s) -
Passamani Giuliana
Publication year - 1989
Publication title -
labour
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.403
H-Index - 34
eISSN - 1467-9914
pISSN - 1121-7081
DOI - 10.1111/j.1467-9914.1989.tb00152.x
Subject(s) - unemployment , duration (music) , economics , term (time) , work (physics) , econometrics , estimation , demographic economics , labour economics , variable (mathematics) , aggregate data , statistics , mathematics , macroeconomics , engineering , mechanical engineering , art , physics , literature , management , quantum mechanics , mathematical analysis
The paper is concerned with estimating and analysing the duration of unemployment, that is the length of time people spend on average looking for work. The first issue of the paper is to estimate unemployment duration using data from the survey on labour force done quarterly in Italy by ISTAT, the National Institute of Statistics. The survey data on the duration of unemployment measure duration so far, that is the average length of unemployment spells in progress up to the date of the survey, but they don't provide any information about completed duration of unemployment experimented by people before finding a job or leaving the labour force. In order to estimate the average length of completed unemployment spells, we would have to use data on cohorts of people followed from the time of entry to the time of exit from the labour market. As longitudinal data is not available, the problem becomes rather complex. One way to get round this is to use data on flows to firstly estimate probabilities of leaving unemployment within a particular period. The available data refer to quarterly flows and yearly flows. This makes it possible to estimate short‐term unemployment (less than six months) and long‐term unemployment (more than twelve months). In another paper we have analysed the nature of the bias introduced by estimating short‐term and long‐term unemployment in the way we do, and we have come to the conclusion that the bias is approximately a constant, which can be very easily estimated and eliminated. The second issue of the paper is to analyse the estimated short‐term and long‐term unemployment in relation to cyclical changes in the economic system and with trend changes in the number of unemployed people seeking the first job. In particular, we want to establish the extent of causal relationships between the chosen explanatory variables and the dependent variable. These causal analyses are done separately for the male and female population, and cover the period from the first quarter of 1979 to the last quarter of 1986.