Premium
A count‐amount model with endogenous recording of observations
Author(s) -
Van Praag B. M. S.,
Vermeulen E. M.
Publication year - 1993
Publication title -
journal of applied econometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.878
H-Index - 99
eISSN - 1099-1255
pISSN - 0883-7252
DOI - 10.1002/jae.3950080406
Subject(s) - econometrics , log normal distribution , value (mathematics) , order (exchange) , distribution (mathematics) , econometric model , event (particle physics) , estimation , economics , statistics , computer science , mathematics , finance , mathematical analysis , physics , management , quantum mechanics
This paper discusses some of the problems which are encountered if an event Ỹ is only recorded if its value satisfies a recording criterion A . It follows that we get an incorrect idea of the frequency of the events and of its true distribution. In order to solve these problems, an econometric model has been constructed by means of which consistent estimation of the true parameters is possible. The model is estimated on consumer purchases, where the number of purchases is assumed to be NEGBIN‐distributed and the purchase amounts obey a lognormal distribution. Purchases are only recorded if their value exceeds Dfl. 10. It is shown that ignoring the recording condition will result in biased estimates and invalid predictions. Apart from this, the model is, among others, relevant for insurance problems, marketing surveys and criminological and epidemiological phenomena.