Mark Y An, Bent Jesper Christensen, and Nabanita Datta Gupta, "Multivariate Mixed Proportional Hazard Modelling of the Joint Retirement of Marries Couples", Journal of Applied Econometrics, Vol. 19, No. 6, 2004, pp. 687-704. There are two data sets, one for the full sample and one for the senior sample. The source of both data sets is Statistics Denmark (http://www.dst.dk). There are 243 observations (couples) in the full sample, and there are 144 observations (couples) in the senior sample. Each dataset is in an ASCII file in DOS format. The two are then zipped in the file acg-data.zip. We describe these two data sets separately. 1. The Full Sample (Section 3.1 of the paper) 1.a The format for the full data file (acg-full.data) is shown using the first three (3) observations: 1 9 81 10 6 0 1 1 1 0 0 agem 52 53 54 55 56 57 58 59 60 61 agef 54 55 56 57 58 59 60 61 62 63 dum84 0 0 0 1 1 1 1 1 1 1 unemp 9.2 9.8 10.5 10.1 9.1 7.9 7.9 8.7 9.5 9.7 2 11 78 8 10 1 0 0 1 0 1 agem 54 55 56 57 58 59 60 61 62 63 agef 47 48 49 50 51 52 53 54 55 56 dum84 0 0 0 0 0 0 1 1 1 1 unemp 7.3 6.1 7 9.2 9.8 10.5 10.1 9.1 7.9 7.9 3 49 72 10 10 1 1 0 1 1 1 agem 54 55 56 57 58 59 60 61 62 63 agef 52 53 54 55 56 57 58 59 60 61 dum84 0 0 0 0 0 0 0 0 0 0 unemp 1.7 1.1 2.5 5.1 5.3 6.4 7.3 6.1 7 9.2 The observations are separated by a blank line. 1.b The first row for each observation is untitled. It lists the values for the following variables (in that order): Obs. No.: from 1 to 243 CoupleID: couple identifier, irrelevant for the analysis Year0: the calendar year that is the base year for the couple Durm: the duration in number of years for the husband Durf: the duration in number of years for the wife Censm: censoring indicator for the husband, =1 if his duration is censored, and =0 if his duration ends with a true retirement. Censf: censoring indicator for the wife, =1 if her duration is censored, and =0 if her duration ends with a true retirement. Skillm: Dummy variable, =1 if husband holds skilled job Skillf: Dummy variable, =1 if wife holds skilled job Owner: Dummy variable, =1 if husband, or wife, or both are homeowner Province: Dummy variable, =1 if the couple lives outside of the capital These variables are all time-invariant. 1.c The next four rows for each observation are the values of the time-varying covariates. The are listed from Year=1 to Year=10. Only the information from Year=1 to Year=Max(durm, durf) is relevant. Beyond that they are filled with the artificial value of 0.00. The four variables are Agem: Age of husband Agef: Age of wife Dum84: Dummy variable for the reform era: year 1984 and after Unemp: Unemployment rate Using the first two, agem and agef, we create other variables to be included in different models for the full sample analysis. 2. The Senior Sample (Section 3.3 of the paper) 2.a The format for the senior sample data file (acg-senior.data) is shown using the first two (2) observations: 1 84 11 2 7 1 0 wealth 31.26 37.81 28.81 31.39 38.85 38.76 44.34 0.00 0.00 0.00 m_inc 18.11 16.52 16.16 15.88 16.09 16.26 16.00 0.00 0.00 0.00 m_pub 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 m_heal 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 m_age67 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 f_inc 10.52 10.69 10.44 10.84 10.83 10.59 10.80 0.00 0.00 0.00 f_pub 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 f_heal 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 f_age60 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 fage 53.00 54.00 55.00 56.00 57.00 58.00 59.00 0.00 0.00 0.00 2 82 51 9 5 0 1 wealth 153.96 141.79 179.22 187.06 192.05 184.63 234.84 198.83 192.15 0.00 m_inc 51.58 46.43 44.57 48.67 48.87 44.73 45.21 37.20 29.65 0.00 m_pub 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.00 m_heal 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 m_age67 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 0.00 f_inc 8.62 10.12 8.95 8.58 4.54 0.00 5.52 9.39 9.62 0.00 f_pub 1.00 1.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 f_heal 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 f_age60 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 0.00 fage 60.00 61.00 62.00 63.00 64.00 65.00 66.00 67.00 68.00 0.00 The observations are separated by a blank line. 2.b The first row for each observation is untitled. It lists the values for the following variables (in that order): Obs. No.: from 1 to 144 Year0: the calendar year that is the base year for the couple CoupleID: couple identifier, irrelevant for the analysis Durm: the duration in number of years for the husband Durf: the duration in number of years for the wife Censm: censoring indicator for the husband, =1 if his duration is censored, and =0 if his duration ends with a true retirement. Censf: censoring indicator for the wife, =1 if her duration is censored, and =0 if her duration ends with a true retirement. These variables are all time-invariant. 2.c The next ten rows for each observation are the values of the time-varying covariates. The are listed from Year=1 to Year=10. Only the information from Year=1 to Year= Max(durm, durf) is relevant. Beyond that they are filled with the artificial value of 0.00. The ten variables are Wealth: couple wealth m_inc: husband's annual income m_pub: Dummy variable, =1 if husband is in public sector m_heal: Dummy variable, =1 if husband is in poor health m_age67: Dummy variable. =1 if husband is eligible for OAP (age 67 and above) f_inc: wife's annual income f_pub: Dummy variable, =1 if wife is in public sector f_heal: Dummy variable, =1 if wife is in poor health f_age60: Dummy variable, =1 if wife is eligible for PEW (age 60 and above) fage: Age of wife