BANDO DI SELEZIONE PER L'ATTRIBUZIONE DI ASSEGNI DI ...

DIPARTIMENTO DI SCIENZE ECONOMICHE
finding something better.
All survey data that try to capture displacement are plagued to some degree by selection
bias. If workers have expectations about the economic viability of their firm then those
workers with good outside prospects in the labor market might quit the firm before it is
closed down or before mass layoffs occur, leaving those colleagues with “worse”
characteristics behind. In restructuring firms that do not close down but initiate mass
layoffs, “better quality” workers might, however, remain with the firm as post-restructuring
productivity gains might imply high wage growth (see e.g. Pfann (2001)). Certain workers
could be encouraged to quit to save the firm redundancy payments. Whatever the selection
mechanism, as long as this mechanism exists, displacement is not a purely exogenous event.
Are selection problems related to mass layoffs and plant closure particularly strong in a
transition economy? Potential failure or poor performance of firms might be easier to
perceive in a transition economy and good workers might then be more likely to leave the
firm long before closure or large-scale labor shedding than in the West. On the other hand,
good workers may have more reason to hold on to their old job in restructuring firms
because of higher future rewards after restructuring. Workers may also hold on to their jobs
because of greater uncertainty in a rapidly changing transition labor market. This
uncertainty is particularly strong in the early stages of transition. Which of these scenarios
prevails in a transition economy is not clear a priori. In this project, we will allow for
unobserved heterogeneity in the estimates of jobless duration and the cost of job loss. At
any rate, when comparing those who quit with those who are displaced, we will keep a
caveat of potential selection bias in mind.
The project will also analyze directly the pecuniary costs of displacement using the method
employed in Lehmann, Philips and Wadsworth (2005). In the Ukrainian case, we have
information on the year and month of any job change and the duration of any intervening
non-employment spell. Respondents are also asked to give gross monthly wages received at
certain periods covered by the survey, typically at one-year intervals. If a worker leaves or
loses a job they are asked to give their final salary. If a worker starts a new job they are
asked to give their starting salary. Those who stay in their job are asked to give their wages
in December of each year and this forms the basis of wage change for the “control” group,
against which wage changes for displaced workers will be compared. Simply comparing
wage changes before and after job loss for displaced workers does not account for any wage
growth that the worker might have experienced had they remained in the job.
The use of a control group of job stayers can account for this effect, (so called “differencein-difference”
estimation). The displacement supplement to the RLFS will have similarly
structured wage questions to ensure comparability across the two countries.
One of the aim of the project will be to collect supplementary information on the predisplacement
wage history of workers, which allows us to get a better handle on an
“Ashenfelter wage dip” that workers might experience even before displacement takes place
as reported by Jacobsen, LaLonde and Sullivan (1993) for U.S. workers. Having predisplacement
wage information also enables us to use a matching variant of the “differencein
difference” estimator that conditions on similar pre-displacement wage profiles of the
treatment (displacement) group and the control group, thus minimizing problems arising
from selection bias. While with national survey data it might be difficult to establish the
“true” pre-displacement wage history of workers, the post-displacement survey data that we