Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/191499 
Year of Publication: 
2017
Series/Report no.: 
IRENE Working Paper No. 17-04
Publisher: 
University of Neuchâtel, Institute of Economic Research (IRENE), Neuchâtel
Abstract: 
We propose in this paper a dynamic n-state transition model to correct for measurement error, that could arise for example from recall and/or design bias, in retrospective panels. Our model allows the correction of measurement errors, when very little auxiliary information is available, over a long period of time taking into consideration the conjuncture fluctuations. The technique suggested shows that it is sufficient to have population moments (for at least one point in time) to correct over- or under-reporting biases. Using a Simulated Method of Moments, one can estimate a transition- and time-specific correction matrix for the labor market flows in a biased retrospective panel. Using retrospective and contemporaneous data from Egypt, we estimate the model and show the significance and robustness of our correction. We show through a reform evaluation that neglecting measurement error in the data would have produced significantly different and misleading results.
Subjects: 
Panel Data
Retrospective Recall
Measurement Error
Labor Markets
Transition Models
JEL: 
C83
C81
J01
J62
J64
Document Type: 
Working Paper

Files in This Item:
File
Size
1.06 MB





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.