Glenn T. Sueyoshi, "A Class of Binary Response Models for Grouped
Duration Data", Journal of Applied Econometrics, Vol. 10, No. 4, 1995,
pp. 411-431.

The data are in a single file called sueyoshi.dat, which is contained in
the zip file sueyoshi.zip.  These files were created on a DOS system, so
the dat file may need to be run through unix2dos or the equivalent if it
is to be used on a Unix system.


1. Data

PSID Layoff Unemployment Spell Data Set Waves 14 and 15 (n=1055 spells)
13,246 binary outcomes

Originally extracted by Larry Katz as reported in his dissertation
and in

Lawrence Katz,  "Layoffs, recall and the duration of unemployment",
NBER Report 1825, 1986.


2. Variables

  1. 1968 ID Number
  2. Recall indicator: 0 = New Job
                       1 = Recall
                       2 = Censored, unemp. at next interview
                       3 = Censored, temp. layoff at next interview
                      10 = Artificially censored at 40
  3. Duration of unemployment spell
  4. Period associated with this binary indicator
  5. Binary indicator for failure in this interval (1=survive; 0=fail)
     The data refer to RECALL survival.  The NEW JOB survival data
     may be extracted from the original Katz data in an analogous
     fashion.
 *6. Age (previous year; year of spell)
 *7. Sex (1 = female)
 *8. Education (years of schooling)
 *9. Race (1 = nonwhite)
*10. Number of dependents
*11. UI dummy (1 = received UI during reported spell, 0 = no UI)
*12. Marital Status (1 = married, previous year)

Occupation Indicators
*13. Laborer, Operative
*14. Craft
*15. Clerical, Services, Sales
*16. Professional and Technical
*17. Manager

Industry Indicators
*18. Metals
*19. Transportation equipment
*20. Other Durable Goods Manufacturing (often in empirical work Metals
     is included in this grouping)
*21. Nondurable Goods Manufacturing
*22. Trade (wholesale & retail)
*23. Transportation and Public Utilities
*24. Mining, Agriculture
*25. Services
*26. Construction

*27. County unemp rate (interval midpoints for 1980 spells)
*28. Wife's employment status (previous interview 1=employed)
*29. Homeowner dummy (1=homeowner)


Variables with *'s have been expressed as deviations from
the original sample means.


3. Format:

Fortran Format Statement

  100  format(5i2/3(9(f8.4)/))

SAS output statement
put id fail durat period synfail2 /
   (sage ssex seduc srace sdepend sui smarry
    slabor scraft scleric sprofess smanage smetals stranseq
    sodurab sndurab strade stransp smining sservice
    sconstr sunemp swife shome) (9*8.4/);