Breakdown Point Theory for Implied Probability Bootstrap
Journal
The Econometrics Journal
ISSN
1368-4221
ISSN-Digital
1368-423X
Type
journal article
Date Issued
2012-02
Author(s)
Camponovo, Lorenzo
;
Otsu, Taisuke
Abstract (De)
This paper studies robustness of bootstrap inference methods under moment conditions. In particular, we compare the uniform weight and implied probability bootstraps by analysing behaviours of the bootstrap quantiles when outliers take arbitrarily large values, and derive the breakdown points for those bootstrap quantiles. The breakdown point properties characterize the situation where the implied probability bootstrap is more robust against outliers than the uniform weight bootstrap. Simulation studies illustrate our theoretical findings.
Language
German
HSG Classification
contribution to scientific community
Refereed
No
Publisher
Wiley-Blackwell
Publisher place
Chichester UK
Volume
15
Number
1
Start page
32
End page
55
Pages
24
Subject(s)
Division(s)
Eprints ID
210468