This is how the model is described in "telegram style". It means that the death rate mijk for a given year j, age i and sex k is specified as mijk = ai fjk , where ai is the effect of the current age and fjk is an interaction term formed from the variables YEAR and SEX.
To assess goodness of fit I computed deviance residuals and plotted them in the form of a Lexis map (not included). Such a presentation of residuals helps to reveal the ages and years that suffer from a lack of fit.
The model specifies that death rate mijk is proportional to the current year, age, and sex mijk = yi aj sk, where yi is the effect of the current year, aj is the effect of the current age, and sk is the effect of sex.
This figure has been produced with the help of the program Lexis. This program is a part of Ph.D. thesis [Andreev 1999] and can be obtained free of charge from Kirill Andreev.
Hereafter if not special references are included author's computations based on WHO mortality database (http://www.who.int/whosis/mort/).