This paper describes a dynamic random effects econometric model from which inferences on earnings mobility may be made. It answers questions such as, given some initial level of observed earnings, what is the probability that an agent with certain characteristics will remain below a specified level of earnings (for example the poverty level) for a specified number of time periods? Existing research assumes that the distributions of the unobserved permanent and transitory shocks in the model are known up to finitely many parameters. However, predictions of earnings mobility are highly sensitive to assumptions about these distributions. The present paper estimates the distributions of the random effects non-parametrically. The results are used to predict the probabilities of remaining in a low state of earnings. The results from the non-parametric distributions are contrasted to those obtained under a normality assumption. Using the non-parametrically estimated distributions gives estimated probabilities that are smaller than those obtained under the normality assumption. Through a Monte Carlo experiment and by examining unconditional predicted earnings distributions, it is demonstrated that the non-parametric method is likely to be considerably more accurate, and that assuming normality may give quite misleading results. Copyright Journal compilation Royal Economic Society 2008. No claim to original US government works
Download Info
To download:
If you experience problems downloading a file, check if you have the
proper application to
view it first. Information about this may be contained
in the File-Format links below. In case of further problems read
the IDEAS help
page. Note that these files are not on the IDEAS
site. Please be patient as the files may be large.
As the access to this document is restricted, you may want to look for a different version under "Related research" (further below) or search for a different version of it.
Volume (Year): 11 (2008) Issue (Month): 3 (November) Pages: 478-498 Download reference. The following formats are available: HTML
(with abstract),
plain text
(with abstract),
BibTeX,
RIS (EndNote, RefMan, ProCite),
ReDIF