The Wage Scar from Youth Unemployment

(2004) - Paul Gregg, Emma Tominey

Journal:
Link::
DOI::
Links::
Tags:: #paper #NCDS #Unemployment #Income #SocialClass
Cite Key:: [@greggWageScarYouth2004]

Abstract

In this paper we utilise the National Child Development Survey to analyse the impact of unemployment during youth upon the wage of individuals up to twenty years later. We find a large and significant wage penalty, even after controlling for educational achievement, region of residence and a wealth of family and individual specific characteristics. We employ an instrumental variables technique to ensure that our results are not driven unobserved individual heterogeneity. Our estimates are robust to the test, indicating that the relationship estimated between youth unemployment and the wage in later life is a causal relationship. Our results suggest a scar from early unemployment in the magnitude of 12% to 15% at age 42. However, this penalty is lower, at 8% to 10%, if individuals avoid repeat incidence of unemployment.

Notes

“We find a large and significant wage penalty, even after controlling for educational achievement, region of residence and a wealth of family and individual specific characteristics.” (Gregg and Tominey, 2004)

“Our estimates are robust to the test, indicating that the relationship estimated between youth unemployment and the wage in later life is a causal relationship.” (Gregg and Tominey, 2004)

“a scar from early unemployment in the magnitude of 12% to 15% at age 42. However, this penalty is lower, at 8% to 10%, if individuals avoid repeat incidence of unemployment.” (Gregg and Tominey, 2004)

“Furthermore, accordin g to Huff Stevens (1997), in the post-displacement period a person is made much more vulnerable to repeated incidence of unemployment.” (Gregg and Tominey, 2004, p. 1)

“This combined with the widening gap between pre- and post- displacement wages in the UK (Nickell et al. 2002) results in long lasting negativ e effects from a spell of unemployment.” (Gregg and Tominey, 2004, p. 1)

“the relationship between early unemployment and later outcomes may not be causal but reflect heterogenei” (Gregg and Tominey, 2004, p. 1)

“If unobservable characteristics of cohort members drive early unemployment experiences and the later wages, our results will be biased upwards. Therefore to ensure the estimated relationship is truly causal we employ the Instrumental Variables technique.” (Gregg and Tominey, 2004, p. 2)

“unemployment rate prevalent locally for individuals aged 16 is employed to instrument youth une mployment in the wage equation for individuals aged 33. The intuition is that at such a young age, the individuals have little autonomy over their area of residence, thus the personal characteristics of the individuals are removed from the equation. Further, the local rate of unemployment certainly plays a role in determining experiences of unemployment. We conclude that unobserved heterogeneity does not create a bias.” (Gregg and Tominey, 2004, p. 2)

“The research in this paper concludes that youth unemployment does indeed impose a wage scar upon individuals, in the magnitude of 12% to 15% at age 42” (Gregg and Tominey, 2004, p. 2)

“In standard unemployment Search Theory, unemployment that is a consequence of an inappropriate match between the employer and employee will have a positive effect on subsequent wages” (Gregg and Tominey, 2004, p. 3)

“Nickel et al. (2002) report for the UK that the cost of job loss rose through the 1980s as wage inequality grew” (Gregg and Tominey, 2004, p. 5)

“Youth unemployment is defined as a period of unemployment covering the ages 16-23” (Gregg and Tominey, 2004, p. 7)

“Heterogeneity is a set of non-time varying observable characteristics of the individual i (Ai) including gender, family background and child ability and unobservable characteristics (Bi), which may capture expectations, aspirations or self -confidence, and an error term (ξit)” (Gregg and Tominey, 2004, p. 8)

“The consequence of failure to take account of heterogeneity is the belief of a strong relationship between unemployment and subsequent wages when, in truth it is not the experience of unemployment per se that results in lower wages, but the unobserved heterogeneity” (Gregg and Tominey, 2004, p. 8)

“This will result in either an omitted variable bias, or a violation of the OLS assumption that the coefficient Uit-1 is correlated with the error term2. Subsequently OLS estimation of Uit-1 will be biased.” (Gregg and Tominey, 2004, p. 8)

“However detailed the set of child and family information available, one or more important background characteristic may have been excluded from the dataset or badly measured. Therefore it is necessary to further control for the unobserved heterogeneity, through econometric techniques. Three main methods are generally adopted in the scarring or cost-of-job-loss literature.” (Gregg and Tominey, 2004, p. 10)

“Difference-in-difference” (Gregg and Tominey, 2004, p. 10)

“Assume functional fo rm” (Gregg and Tominey, 2004, p. 10)

“Instrumental variables” (Gregg and Tominey, 2004, p. 11)

“To avoid the requirement of non-linear instruments, a linear model is adopted, where the effect of one month of youth unemployment upon the natural log of the hourly wage reported at age 33 is evaluated.” (Gregg and Tominey, 2004, p. 22)

“One month of unemployment will reduce a male individual’s wage at 33 by 0.8% and will reduce a female individual’s wage at 33 by 0.7%, conditional on education level, measures of family background and ability and ward level unemployment at age 33 from the 1991 Census. The application of the IV technique does change the coefficients; the estimated impact of months of youth unemployment upon the subsequent wage rises slightly, although the results are not largely different from the OLS estimates.” (Gregg and Tominey, 2004, p. 22)

“Our findings are that youth unemployment imposes a sizeable wage scar upon both males and females at age 23 followed by substantial recovery over the next ten years, but only if the individual can avoid further spells of unemployment after age 23. A modest residual wage scar of around 8% persists up to twenty years later even for those who have no further unemployment experience. Those with extensive youth unemployment are at higher risk of further unemployment through to age 33 and this inhibits wage recovery. However there was no further relationship between youth unemployment and unemployment reported after age 33.” (Gregg and Tominey, 2004, p. 23)