Lorenzo Cappellari and Stephen P. Jenkins, "Modelling Low Income Transitions", Journal of Applied Econometrics, Vol. 19, No. 5, 2004, pp. 593-610. The paper uses data from the British Household Panel Survey (BHPS): (i) the official public release files, plus (ii) the unofficial 'Net income variables to accompany BHPS waves 1-9' files. Bona fide researchers may access (i) and (ii) by registering with the UK Data Archive and thereby also undertaking to abide by various rules concerning maintenance of data confidentiality. To register with the Data Archive, and to browse documentation, visit their website: http://www.data-archive.ac.uk. We used data from waves 1-9 (survey years 1991-1999) of the BHPS. Low income transitions refer to transitions between two consecutive waves (years 't-1' and 't'). Transitions from the panel were pooled; a given individual could therefore contribute between one and eight transitions. The estimation sample was restricted to individuals aged 20-59 years in year t-1 who were not in full-time education. Each individual's income was defined to be the post-tax post-transfer income of the household to which he or she belonged, adjusted for differences in household size and composition using the 'McClements' equivalence scale, and expressed in August 2000 prices. (See the documentation for (ii) for further details.) An individual was defined to have a low income at t if he or she had an income below 60% of median income at t. (Several alternative low-income cut-offs were also used, as mentioned in the paper.) We used the following BHPS variables. (Note that, in the public release files, variable names are prefixed by a letter indicating the survey wave, e.g. 'a' for wave 1, 'b' for wave 2, and so on. These prefixes are omitted below.) pid Person identifier hid Household identifier hhneti Post-tax post-transfer current household income age Age hgsex Gender qfachi Highest academic qualification jbhrs Number of hours normally worked per week jshrs Number of hours normally worked per week (self-employed) race Ethnic group hhtype Household type jbsemp Employment status tenure Housing tenure butype Benefit unit type maju Mother not working when respondent aged 14 maju Father not working when respondent aged 14 masoc Mother's occupation (SOC), respondent aged 14 pasoc Father's occupation (SOC), respondent aged 14 hgr2r Relationship to household reference person The Stata .do file code that was used to extract the estimation sample and to construct derived variables is available from the authors upon request. Lorenzo Cappellari and Stephen P. Jenkins