Dümbgen, Lutz (2016). (Ab)Using Regression for Data Adjustment (Technical report 78). Bern: IMSV, University of Bern

Text
AdjustmentsRegression.pdf  Published Version Available under License BORIS Standard License. Download (681kB)  Preview 
In various economic applications, people want to compare $n$ units with respect to certain quantities Y₁, Y₂, ..., Yn measuring their performance. The latter, however, is often influenced by certain factors which are beyond control of the units, and one would like to extract an adjusted performance from the data. Specifically, let Xi ⋲ X summarize the factors of the ith unit. Then one could think of a model equation Yi = fo(Xi) + ɛi with a regression function fo : X → R describing the unavoidable influence of the factors Xi and ɛi being the adjusted performance of the ith unit. Now a common proposal is to estimate fo via regression methods by a function ^f depending on the current data (Xi,Yi), possibly augmented by additional past data, and to use the residuals ^ɛi := Yi  ^f(Xi) as surrogates for the adjusted performances ɛi. In the present report we discuss this approach, its potential pitfalls and (mis)interpretation. In particular, an unavoidable property of the residuals ^ɛi$ is that they measure only parts of the adjusted performance while the remaining parts get hidden in the estimated function ^f. Possible alternatives are mentioned briefly.
Item Type: 
Report (Report) 

Division/Institute: 
08 Faculty of Science > Department of Mathematics and Statistics > Institute of Mathematical Statistics and Actuarial Science 
UniBE Contributor: 
Dümbgen, Lutz 
Subjects: 
300 Social sciences, sociology & anthropology > 330 Economics 500 Science > 510 Mathematics 
Series: 
Technical report 
Publisher: 
IMSV, University of Bern 
Language: 
English 
Submitter: 
Lutz Dümbgen 
Date Deposited: 
05 Sep 2016 10:57 
Last Modified: 
10 Sep 2017 15:30 
ArXiv ID: 
1202.1964 
BORIS DOI: 
10.7892/boris.87395 
URI: 
https://boris.unibe.ch/id/eprint/87395 