• UCL Twitter account
  • UCL YouTube channel
  • UCL Facebook page
  • UCL SoundCloud channel
  • UCL iTunes store

Information for Staff

Calendar

Select dates to view past and future events

12:30 - 13:30 27 November 2012

Estimation with Contaminated Data

Location

IFS Seminar Room | Institute for Fiscal Studies (link Map)
7 Ridgmount Street | London | WC1E 7AE |

Open to: Academic | Student
Ticketing: Open

Speaker information

David Pacini, Bristol

This paper studies the problem of identification and estimation of unknown parameters of interest defined by moment equality restrictions when data are available from a contaminated sample. This problem arises, for instance, when researchers try to estimate linear instrumental variable models from sample data with outliers. The challenge is to construct an estimator of the parameter of interest reflecting both contamination and sampling variability. Drawing on earlier work on set identification of distribution functions from contaminated data, we derive a simulated modified method-of-moments set estimator. The proposed set estimator permits to report the effects of contamination and sampling variability in situations where existing results do not directly apply, such as linear projections, quantile linear regressions, or linear instrumental variable models with contaminated covariates. We the proposed set estimator with existing robust point estimators via Monte Carlo exercises.


Contact

Institute for Fiscal Studies
020 7291 4800 |


image