INFLUENCES OF PAST HISTORY ON THE INCIDENCE OF YOUTH UNEMPLOYMENT: EMPIRICAL FINDINGS FOR THE UK†
DOI | http://doi.org/10.1111/j.1468-0084.1993.mp55002002.x |
Author | Peter Elias,Wiji Narendranathan |
Published date | 01 May 1993 |
Date | 01 May 1993 |
OXFORD BULLETIN OF ECONOMICS AND STATISTICS, 55,2(1993)
0305-9049 $3.00
INFLUENCES OF PAST HISTORY ON THE
INCIDENCE OF YOUTH UNEMPLOYMENT:
EMPIRICAL FINDINGS FOR THE UKt
W&i Narendranathan and Peter Elias
I. INTRODUCTION
The issue of whether or not there is evidence of a causal relationship between
the experience of unemployment and the future economic activity of an
individual is, as yet, unresolved by labour economists. Theoretical reasoning
suggests that one may expect to find such a relationship, either through a
reduction in human capital or through employer recruitment and labour
retention practices. The use of a 'last-in first-out' policy to shed workers can
imply a causal relationship between past and future spells of unemployment
for the individual. Equally, a reduction in voluntary job-to-job mobility
associated with increased tenure with an employer puts those employees with
longer tenure at a reduced risk of experiencing unemployment.
The importance of this issue, in terms of its implications for policies which
address youth labour markets, should not be underestimated. Young people
demonstrate high rates of labour turnover, relative to other age groups, for
reasons related to their 'sampling' of different types of work and different
employers, their low acquisition of employer or occupation-specific human
capital and their relative lack of constraints associated with family formation.
But high turnover carries with it a risk of unemployment. If the experience of
unemployment itself increases the risk of future unemployment, the resulting
concentration of unemployment can have serious repercussions as young
people enter early adulthood.
To address these issues, we need longitudina[ information on the time
order of spells of unemployment, their incidence and duration. As discussed
in detail in the next section, the longitudinal information which we utilize in
tWe are grateful to Chris Jones for her excellent programming assistance and to John
Micklewright, Steve Nickell and Jonathan Thomas for helpful comments. Comments received
from the participants of the seminars held at the 6th World Congress of the Econometric
Society, Barcelona, Spain in August 1990, Australian National University, Monash University
and University of Tasmania, and from participants at a conference on 'Recent developments in
unemployment' held at the Institute for Employment Research, University of Warwick in
November 1990, are also gratefully acknowledged.
161
162 BULLETIN
this study is collected as monthly records of labour market status. Short spells
which lasted for less than two weeks were not recorded, neither was the
actual completion date of a spell. Given these shortcomings, it did not seem
appropriate to use cóntinuous time models to address the issues under
consideration in this paper using these types of longitudinal data. Researchers
who had more detailed information on the duration of unemployment for a
cross-section of individuals have thus focused mainly upon the issue of
duration dependence in spells of unemployment' and based their analyses on
continuous time models. Ellwood (1982) who based his study upon US data,
found no evidence of a negative relationship between the duration of a spell
of unemployment and the probability of exit from unemployment. However,
using a larger and more detailed data set, Lynch (1989) concluded that ... as
the spell of non-employment increases, the probability of becoming re-
employed declines sharply' (p. 45). In her similar study of the unemployment
durations among a sample of 68 youths from Inner London boroughs, Lynch
(1985) reaches the same conclusion. Based upon monthly labour market
history data of young men who just graduated from high school in the US,
Heckman and Borjas (1980), using a continuous time model, found no
evidence that previous occurrences of unemployment or their duration affect
future labour market behaviour once sample selection and omitted hetero-
geneity biases were controlled for in their analysis. Although the question of
whether or not there is duration dependence in the exit probability from a
relatively short spell of unemployment is of intrinsic interest, the issue of
whether or not a long term pattern of recurrent unemployment begins to
develop in the early years of a person's working life is also very important.
This paper considers the issues which lend complexity to this problem and
seeks to address these issues using a variety of techniques. In particular,
problems arising from interval and point sampling of longitudinal informa-
tion, the effects of observed and unobserved heterogeneity and from the lack
of control for serially correlated exogenous factors are investigated. By focus-
ing upon the seven-year work histories of a group of young males, all of
whom left school in 1974 at the age of 16 years, this study also enables us to
address the exogeneity/endogeneity issue raised by the initial status of these
individuals.
The plan of the paper is as follows: the next section gives a brief introduc-
tion to the data set analysed. Section III sets out the model, discusses the
different estimation techniques used in the analysis and addresses some
important econometric issues raised by this type of analysis. A detailed
description of the sample used and the main findings are presented in Section
IV and the paper summarizes and concludes in Section V.
'In all of these studies, the unemployment duration is modelled by specifying the conditional
probability of leaving unemployment (the hazard function) without distinguishing the exit state.
However, some of those who exit the unemployment do not do so into employment. Thus, one
should strictly interpret the evidences as being about the exit probability from unemployment
rather than as being about the re-employment probability.
To continue reading
Request your trial