Fabrizia Mealli and Stephen Pudney, "Occupational Pensions and Job
Mobility in Britain: Estimation of a Random-Effects Competing Risks
Model", Journal of Applied Econometrics, Vol. 11, No. 3, 1996, pp.
293-320. 

AUTHORS:  Fabrizia Mealli (Department of Statistics, University of Florence)
          Stephen Pudney (University of Leicester)

SOURCE OF DATA:  
    UK Retirement and Retirement Plans Survey 1988/89
    (see M.Bone, J.Gregory, B.Gill and D.Lader (OPCS) "Retirement and 
    retirement plans", published by HMSO 1992, ISBN 0 11 691386 X)

ORGANISATION OF DATA:  
    The data are in two flat files in ASCII format, delimited by blanks.
    Both files (which are in DOS format) are zipped in the file mp-data.zip.

File INDIV.ASC contains variables describing the characteristics of the
individuals included in our subsample; there are 1334 records, each
containing an observation on the following five variables: 

(1) IDENTIFIER - numerical code running from 1 to 1334;

(2) NUMBER OF SPELLS - the number of separate episodes observed for the 
                       individual;
(3) INITIAL STATE - 
            code identifying the state in which the first episode is spent:
            1 = pensioned job; 2 = non-pensioned job; 3 = other employment; 
            4 = unemployment; 5 = out of the labour force;
(4) AGEFTE - age at which (continuous) full-time education finished;

(5) DEGREE - dummy identifying those with a university degree

File EPISODE.ASC contains variables describing individual episodes; there
are 3578 records, each describing a single episode. The number of records
relating to each individual is given by the second variable in file
INDIV.ASC. The blocks of records relating to different individuals are
ordered in the same way as individuals are ordered in INDIV.ASC. There are
10 variables describing each episode: 

(1) DURATION - duration of the episode in years;

(2) EXIT ROUTE - discrete indicator specifying how the episode ended:
     1 ... 5 = transition to state 1 .. 5 (P, NP, OE, U, OLF); or
         -99 = censored episode (note that the final episode for an individual 
               is always censored);

(3) P DUMMY - dummy variable equal to 1 if the current episode is spent in
              state P (pensioned job); 

(4) NP DUMMY - dummy variable equal to 1 if the current episode is spent
               in state NP (non-pensioned job);

(5) OE DUMMY - dummy variable equal to 1 if the current episode is spent
               in state OE (other employment);

(6) U DUMMY - dummy variable equal to 1 if the current episode is spent in
              state U (unemployment);

(7) MARRIED - dummy equal to 1 if individual is married at start of episode;

(8) AGE - age in years at start of the episode;

(9) DATE - date of the start of the episode (years since 1900);

(10) TIME OUT - time (in years) spent out of the labour force.